Farseer-NMR: automatic treatment, analysis and plotting of large, multi-variable NMR data

Farseer-NMR: automatic treatment, analysis and plotting of large, multi-variable NMR data We present Farseer-NMR (https://git.io/vA ueU), a sof tware package to treat, evaluate and combine NMR spectroscopic data from sets of protein-derived peaklists covering a range of experimental conditions. The combined advances in NMR and molecular biology enable the study of complex biomolecular systems such as flexible proteins or large multibody complexes, which display a strong and functionally relevant response to their environmental conditions, e.g. the presence of ligands, site- directed mutations, post translational modifications, molecular crowders or the chemical composition of the solution. These advances have created a growing need to analyse those systems’ responses to multiple variables. The combined analysis of NMR peaklists from large and multivariable datasets has become a new bottleneck in the NMR analysis pipeline, whereby information-rich NMR-derived parameters have to be manually generated, which can be tedious, repetitive and prone to human error, or even unfeasible for very large datasets. There is a persistent gap in the development and distribution of soft- ware focused on peaklist treatment, analysis and representation, and specifically able to handle large multivariable datasets, which are becoming more commonplace. In this regard, Farseer-NMR aims to close this longstanding gap in the automated NMR user pipeline and, altogether, reduce the time burden of analysis of large sets of peaklists from days/weeks to seconds/ minutes. We have implemented some of the most common, as well as new, routines for calculation of NMR parameters and several publication-quality plotting templates to improve NMR data representation. Farseer-NMR has been written entirely in Python and its modular code base enables facile extension. Keywords NMR spectroscopy · Data analysis · Intrinsically disordered proteins · Paramagnetic-NMR · Proteins · Chemical shift perturbations Introduction Electronic supplementary material The online version of this article (https ://doi.org/10.1007/s1085 8-018-0182-5) contains Flexible proteins, including fully disordered proteins, short supplementary material, which is available to authorized users. disordered segments or interdomain linkers, as well as multi- body macromolecular complexes, play a fundamental role * João M. C. Teixeira in many key regulatory processes by which environmental joaomcteixeira@gmail.com information is sensed, processed or used to modulate com- Simon P. Skinner plex responses. The reaction of these systems to environ- s.p.skinner@leeds.ac.uk mental changes, which provide fundamental information on Miguel Arbesú their function, can be experimentally investigated by explor- miguelarbesu@ub.edu ing them under a large set of conditions and evaluating their Alexander L. Breeze multivariable response. In this regard, Nuclear magnetic res- a.l.breeze@leeds.ac.uk onance (NMR) generates atomic level data bearing structural Miquel Pons and dynamic information. Over the last two decades, the mpons@ub.edu information that can be extracted from NMR observables BioNMR Group, Inorganic and Organic Chemistry has continuously increased owing to the development of Department, University of Barcelona, Barcelona, Spain sophisticated NMR methods combined with molecular biol- Astbury Centre for Structural Molecular Biology, Faculty ogy protocols for sample preparation, which together have of Biological Sciences, University of Leeds, Leeds, UK Vol.:(0123456789) 1 3 2 Journal of Biomolecular NMR (2018) 71:1–9 considerably enriched the NMR spectroscopist’s toolkit. The NMR-peaklist text files which contain the NMR observa- advent of these advances places NMR in the front line as bles. Moreover, there are multiple ways of combining the one of the most valuable techniques to investigate complex variable-dependent information, increasing the complex- protein systems under a multitude of conditions (Kay 2016; ity of the analysis and leading to an overload of repetitive Sormanni et al. 2017), which include the mapping of ligand tasks if peaklists are treated manually. binding sites and their affinities (Arai et al. 2012; Teilum In addition to the above arguments, the nature of NMR et al. 2017), the presence of molecular crowders (Diniz et al. users, and consequently NMR software users, has also 2017), different chemical compositions of the solution (Sen- evolved over the past two decades, from specialist NMR gupta et al. 2017) as well as site-directed mutations (Arbesú spectroscopists to structural biologists, who are not nec- et al. 2017) and local post-translational modifications (Theil- essarily NMR experts, but use the technique routinely for let et al. 2012) whose effects expand to the whole molecule. data acquisition and analysis. As a result of the evolution Such experimental design generates large and multivariable of the biomolecular NMR user base, there is an increasing sets of NMR data. Complex datasets are also generated in need for the development of a user-directed NMR pipeline, the context of time-resolved NMR studies (Theillet et al. which focuses on the needs of the community as a whole. 2013; Mylona et al. 2016), and structural/dynamic analy- The interplay between NMR and other techniques for struc- sis by paramagnetic observables (Mahawaththa et al. 2017) tural biology is accelerating this paradigm shift (Wassenaar from different sources (Carlon et al. 2016; Schilder et al. et al. 2012, http://www.inext -eu.org). Moreover, we believe 2016). Nonetheless, exploiting a combined use of such avail- that such a pipeline should be open, easily extensible and able NMR tools is not yet routine, mostly due to a lack of accessible to anyone who wants to use and contribute to it, platforms that close the gap between the expertise of devel- in compliance with the most up-to-date Open Access poli- opers and the users’ needs, both at the acquisition and the cies of the European Community (http://ec.europ a.eu/resea analysis level.rch/openscienc e/inde x.cfm ), and not closed to all except the Numerous NMR data analysis packages have been software’s developers. developed over the past 25 years (Maciejewski et al. 2017); Taken together, the above arguments have motivated us this includes packages for specific applications e.g. relaxa- to develop Farseer-NMR, a user-directed, open source and tion analysis (Mandel et al. 1995; Orekhov et al. 1995; modular toolbox into which users can input and analyse d’Auvergne and Gooley 2008), structure solution (Güntert large datasets of multivariable NMR curated peaklists in an et  al. 1997; Herrmann et al. 2002; Rieping et al. 2007; efficient, reproducible and organized manner and without Güntert 2009), paramagnetic NMR analysis (Pintacuda a steep learning curve. The current version has been spe- et al. 2004; Schmitz et al. 2006, 2008; John et al. 2007; cifically designed for the analysis of protein-related data. Skinner et al. 2013), resonance assignment (Zimmerman Farseer-NMR transforms the analysis of numerous peaklists et al. 1997; Jung and Zweckstetter 2004a, b; Narayanan from weeks of work to a few minutes of automatic execution, et  al. 2010), and comprehensive packages to perform a as is represented in Fig. 1 of Farseer-NMR Documentation multitude of different analyses mainly focused on spectral provided in Supplementary Material, something which has processing and data representation, whilst also taking care been lacking in the NMR analysis pipeline. The software is of the associated bookkeeping (Johnson and Blevins 1994; coded using the Python programming language. We pro- Bartels et  al. 1995; Delaglio et  al. 1995; Keller 2003; pose Farseer-NMR also as a nucleation point for the vari- Vranken et al. 2005; Lee et al. 2015; Skinner et al. 2016). ous NMR software packages/routines built to treat NMR- In spite of the plethora of analysis that these packages can derived parameters, in particular those without a graphical perform, which unarguably constitute milestone achieve- user interface (GUI). We have endeavoured to facilitate this ments in the field, we persistently find a gap on software by providing a platform upon which additional routines can availability dedicated to conversion of NMR observables be developed and launched. Farseer-NMR has been devel- in large sets of peaklists into correlated information-rich oped around the concept of a “transparent box”, rather than a parameters from which biologically relevant answers can “black box”, or just another Python library for programmers be extracted and comfortably presented to the community. to use. In this respect, we distribute Farseer-NMR with a This gap expands as the complexity of the systems under complete documentation record which, together with inline the scope of biomolecular NMR projects increases and code commentaries, explain the program structure and archi- the number of variables needed to characterise them also tecture, provide examples of new implementations, both grows. Hence, it is often necessary to acquire a multitude for core and for the user interface, and describe the setup of NMR spectra covering this whole range of variables— and execution of a Farseer-NMR calculation. All settings we term these datasets multivariable. Upon spectral analy- required for a Farseer-NMR calculation are user-settable, sis, the researcher faces the task of extracting structural either using a configuration file, or via the GUI. This report information from correlated series of, maybe hundreds, describes the concepts of Farseer-NMR, its data structure 1 3 Journal of Biomolecular NMR (2018) 71:1–9 3 alpha-release, it is fully operational, but very much open to improvements that will be directed by the interaction with the community. The artwork in this manuscript was prepared using the Matplotlib functions developed for Farseer-NMR, the LibreOffice Suite (https ://www.libre offic e.org/), Adobe Photoshop CS2 (http://www .adobe .com ) and assembled with GIMP (https ://www.gimp.org/). The Farseer‑NMR structure Farseer-NMR has been developed to enable analysis of mul- tivariable sets of protein-related NMR data in a concerted, correlated and automated manner; variables can be continu- ous e.g. concentration of a titrated ligand, or discreet e.g. dif- ferent protein constructs, ligands, environment. User curated peaklists from two-dimensional NMR spectra are imported Fig. 1 A schematic representation of the Farseer-NMR Cube. Exam- ples of different variables are given of each Cube’s axis. Each col- into the Farseer-NMR data structure, and from here, experi- oured rectangle, the first one highlighted with ‘.csv’ for representa- mental datasets can be freely navigated and exploited for tion is a 2D-NMR peaklist analysis without any restrictions of the acquisition schedule. Up to three experimental variables can be simultaneously and implementation of its functionalities, all of which do analysed by Farseer-NMR and, therefore, we refer to the not require any programming knowledge. data structure as the “Farseer-NMR cube” (Fig. 1 and Sect. II-a of the Documentation PDF), on which each dimension/ axis, x, y or z, of the cube represents a different experimen- Methods tal variable and each cube data point, a coordinate within these axes, is an NMR peaklist table composed of rows that The Farseer-NMR Project is hosted at GitHub (https://git hu represent the protein residues and columns that contain all b.com) and distributed under the GPL-3.0 license (https :// residue-related information extracted from the NMR spectra www.fsf.org/); the following link directs to the project front alongside user annotations. This configuration yields a final page: https://git.io/vA ueU. F arseer-NMR was written using 5D matrix-like structure stored in a single digital object. version 3.6 of the Python programming language, and vari- The Farseer-NMR Cube can be traversed along the different ous Python libraries were used for data management and dimensions describing the dependency of the data on the analysis. Peaklist treatment and management, parameter different defined variables. calculation and data fitting are performed using a combi - For example, if a data set contains diamagnetic and para- nation of routines from Numpy and Scipy (van der Walt magnetic spectra of two different protein constructs, into et al. 2011) and Pandas (Wes 2010), and plotting templates which a ligand concentration range was titrated, Farseer- were developed using Matplotlib (Hunter 2007). The GUI NMR can analyse the spectral variations (for example, chem- was developed using PyQt v5.7. Version control of the code ical shifts, intensities, linewidths, couplings) with respect base was achieved using a git repository hosted by GitHub to ligand concentration and for each construct separately (http://www.github.com ). Farseer-NMR was developed and (extracting information that can be used to calculate binding tested using the GNU/Linux operating system and Anaconda affinities, x axis), calculate the changes caused by a given Python (http://www.anaco nda.com) distribution, addition- ligand concentration between the two constructs (y axis), ally we provide Anaconda environment files for users to and extract paramagnetic relaxation enhancements (PRE) setup a running environment if necessary. We have also for all spectra by comparing the paramagnetic series with produced complete documentation containing a plethora the diamagnetic one (z axis), from which distance dependent of technical information about Farseer-NMR, its concepts information can be extracted. Additionally, data calculated and its functionalities. A maintained version of the Farseer- along different axes can be parsed and represented together NMR Documentation is available online from the home page along the other axes (comparative analysis, see Case Study). repository, the version corresponding to the current release The selection of different dimensions of the cube for analysis has been supplied as Supplementary Material. At the time and comparison across dimensions are Boolean flags in the of this manuscript, the version of Farseer-NMR is that of an setup of a Farseer-NMR calculation (vide infra). Axes can 1 3 4 Journal of Biomolecular NMR (2018) 71:1–9 have any number of data points (peaklists) without compro- Farseer‑NMR analysis routines mising the organization and representation of results. Any type of data can be specified along any of the differ - Each series extracted from the Farseer-NMR cube can be ent x, y, and z axes; however, for the sake of simplicity and analysed using a number of analysis routines, which are clarity, some analysis routines are currently restricted to spe- user-configurable. Some commonly used calculation and cific axes, e.g. paramagnetic analysis can only be performed plotting routines for NMR analysis of macromolecular data along the z axis, inclusion of different protein sequences can have been implemented in the Farseer-NMR code base; only be performed along the y axis, and fitting of continu - we have also included novel methods/templates among ous data can only be performed along the x axis. Complete which is the recently developed ΔPRE method (Arbesú descriptions of axis characteristics can be found in Sect. V-d et al. 2017). In addition, and more importantly, the core of the Farseer-NMR Documentation. of Farseer-NMR has been designed such that virtually any calculation or plotting routine can be implemented. Far- seer-NMR analysis routines consist of three steps, namely, The Farseer‑NMR workflow (i) calculation of NMR parameters from raw data, (ii) data and values export and (iii) plotting. The current version One of the great advantages of having all experimental data of Farseer-NMR can calculate nucleus-specific and com- in one computational object, the Farseer-NMR Cube, is the bined chemical shift perturbations (CSPs), according to ability to slice the cube in any direction and interrogate the Williamson (2013), height and volume ratios, and ΔPRE data to investigate specific questions or phenomena which values according to Arbesú et al. (2017). Elaborated NMR are not limited to the acquisition schedule. The first step parameters are calculated by evaluating each peaklist in an of any Farseer-NMR calculation is to load the user curated experimental series with the appropriate reference experi- peaklists as digital tables. Farseer-NMR accepts as input ment, which is the first in the series, and the calculated peaklists from a variety of formats, namely, Ansig, CCPN values are stored in the corresponding peaklist in newly Analysis version 2 (Vranken et al. 2005), NMRPipe (Dela- added columns. Farseer-NMR also contains a restraint fit- glio et al. 1995), NmrView (Johnson and Blevins 1994) and ting platform, whereby continuous data can be fit, cur - Sparky (Lee et al. 2015), taking a step towards the integra- rently restricted to data contained along the x axis. It is not tion of the NMR community as a whole. Once imported, the aim of Farseer-NMR to provide, presently, a complete the peaklists are converted into an enhanced format initially suite of routines for fitting the most diverse natures of derived from CCPN Analysis version 2 and are stored in a continuous NMR data, instead, to provide a platform for variable-defined hierarchical directory structure created by which users/developers can implement their own fitting Farseer-NMR. The peaklists dataset is loaded to memory as routines that can plug into the core Farseer-NMR routines, a nested dictionary. At this stage, peaklists are scanned for and, in this way, open the door for an easy distribution identification of missing and/or unassigned residues along among the whole NMR community. the first variable (x axis) and new rows representing these During the calculation run, a set of folders, organised residues are added to the peaklists. This serves two pur- hierarchically according to the defined variables, is cre- poses: identification of those residues and normalization of ated to store all Farseer-NMR output, so that the user can the peaklists to the same number of rows; these procedures readily access the generated results. Within these folders, are described in detail in Sect. II-c of the Farseer-NMR formatted, treated and parsed peaklists are exported as Documentation. Amide sidechain entries are also identified comma-separated files, which contain the newly calcu- at this stage, if required. Following peaklist treatment, the lated NMR parameters in addition to the original data. Farseer-NMR Cube is generated. Moreover, the user-specified plots are created alongside All settings required by Farseer-NMR for a calculation: parsed text tables containing only the represented data. selection of dimensions for analysis, types of analysis, types UCSF Chimera (Pettersen et al. 2004) attribute files are of output plots and settings for the individual plot types, generated to allow straightforward representation of the are specified in a JSON file (http://www.json.org/), which calculated restraints in 3D molecular structures; facile can be prepared manually or via the GUI (vide infra). Once extension of this feature to meet other molecular repre- executed, Farseer-NMR sequentially generates series of sentation software requirements is possible. At present, peaklists from the Farseer-NMR cube, which represent Farseer-NMR contains eight plotting templates (three bar the system’s evolution along an axis/variable at fixed data plots, two scatter plots, a heat map and two continuous points of the other two axes and walks through all the pos- data plots), which have been prepared specifically for sible combinations of x, y and z variables, (Sect. II-b of NMR-derived data representation. These represent some the Farseer-NMR Documentation) and analyses these series of the most common plotting styles in NMR literature, according to the user-specified settings. 1 3 Journal of Biomolecular NMR (2018) 71:1–9 5 along with novel methods of presenting NMR-derived The graphical user interface (GUI) data. The plots generated by Farseer-NMR are composed of subplots that either represent data for all residues in To facilitate easy preparation and execution of Farseer-NMR a given series, or residue-specific variations for a given calculation runs, a user-friendly GUI has been developed; variable (where appropriate). This enables facile identi- full description can be found in Sect. III of the Farseer-NMR fication of significant results from a single figure, either Documentation. The current GUI is divided into two tabs, by condition or by residue. Default configurations for all namely, Peaklist Selection (Fig. 3) and Settings (Sect. III-c plots have been implemented, which yield publication- of the Documentation PDF). In similarity with the core code, quality figures, although templates can be fully config- the GUI was developed to be easily expansible and designed ured by the user. Figure  2 shows examples of three of so that new features can be added as new tabs with their the plotting templates while full information regarding respective options and functions. We envisage user-devel- Farseer-NMR plotting features is available in Sect. VI-d oper contributions playing a significant role in this evolution of the Farseer-NMR Documentation. If continuous data of the interface. The Peaklist Selection tab consists of three fitting is performed as part of a Farseer-NMR calculation distinct areas: a sidebar for data import and selection, an run, the values obtained from fitting can be represented as area for specifying experimental conditions, and an area for a plot of restraint evolution per residue. Farseer-NMR has construction of the Farseer-NMR data structure. Peaklists already been used to generate figures in published work are imported by dropping them on to the side bar, which will (Bijlmakers et  al. 2016; Marimon et  al. 2016; Arbesú detect the peaklist format and import the data if the peaklist et al. 2017). format is recognised (vide supra). Setting up the Farseer- NMR data structure consists of two stages: (1) specifying the numbers of points and labels for each of the axes of the Far- seer-NMR cube, that is, experimental variables and number Fig. 2 Three examples of the implemented Farseer-NMR plotting lines are identified by character “P”. Blue and gold bars (labelled 1 1 15 templates. H and N chemical shifts were generated from a syn- and 2, respectively) represent two pairs of residues with alternative thetic data set simulating seven points of a ligand titration of a 100 assignment, that after plot analysis can be swapped with confidence residues protein. a Extended bar plot template and b compacted bar (14F ↔ 48R, 18Q ↔ 97M), colours and labels are representations plot template representing the combined chemical shift perturbations of the user annotations present in the original peaklists. A signifi- (CSP), calculated according to Williamson (2013), for the last experi- cance threshold is represented as a red line. c Progression of the CSP ment in the series (with 800 µM ligand) versus the reference experi- parameter represented individually for each residue. Peaks that have ment (ligand free). Black bars represent residues measured in that disappeared along the series are easily identified (22F), as well as spectrum. Unassigned residues are represented in grey a × ticks or b unassigned peaks. The represented plots are crops of the full pictures background shade. Red bars represent missing residues, i.e. residues generated by Farseer-NMR that represent the whole series. All col- that have been observed previously in the series but have disappeared ours, sizes, shapes, and labels are user-configurable at a given ligand concentration; the last measured value is kept. Pro- 1 3 6 Journal of Biomolecular NMR (2018) 71:1–9 Fig. 3 A screenshot of the Farseer-NMR user interface. The Peaklist selection tab is shown loaded with an artificial dataset as an example. The Farseer-NMR Cube variables are set up and the corresponding data points are populated with peaklists of measured data points, respectively, and (2) selection of setting menus under a new tab. Default settings for all of the appropriate peaklist for each combination of variables. the fields, are set at start-up, and configuration files can also Once the numbers of points, and labels for each axis point be loaded and saved from this tab. The footer of the GUI have been specified, a hierarchical tree diagram representing enables instant access to Farseer-NMR documentation, to all specified conditions is drawn with “Drop peaklist here”, the Farseer-NMR Twitter account, and to contact the devel- written at the end of each branch. From here, peaklists cor- opers via e-mail to submit feedback, request features and responding to the correct combination of variables can be report any bugs. dragged and dropped onto the end of each branch of the tree, thereby creating the Farseer-NMR data structure. The Settings tab of the GUI contains input fields for all Case study: ΔPRE analysis the parameters needed to analyse the data using Farseer- NMR. There are two text input boxes, one for setting An example of the versatility of Farseer-NMR is the imple- the Peaklist Dataset Folder from which peaklists can be mentation of a new procedure for the analysis of intramo- imported aside from the drag-and-drop area, and a second lecular PRE induced by a paramagnetic probe on disordered for setting the Calculation Output Path to which all treated protein fragments of a series of proteins harboring mutations peaklists and plots will be written. Selection of axes along that may affect their dynamics and conformational properties which to perform analysis, and options to compare across (Arbesú et al 2017). The data set was formed of 10 sequence dimensions, consider side chain amide peaks and identifi- variants that probed the effect of four point mutations at cation of missing and unassigned residues, are specified by conserved aromatic residues and the presence of the folded checkboxes in this tab. Settings for chemical shift normalisa- domain adjacent to the disordered region by measuring the tion, NMR observable selection and parameter calculation, relaxation induced by paramagnetic centre incorporated in plotting template selection, figure format and file dimension three alternative sites along the sequence. Experimental data and resolution are also configurable from this tab. A series consisted of HSQC spectra of the paramagnetic proteins and of popups accessible from this tab contain all the settings the corresponding diamagnetic control (20 experiments). for all the implemented plotting templates. The GUI has The observed PREs were compared with the values expected been forged to facilitate, and invite, advanced users to imple- from a random coil model for the disordered region, inde- ment new calculation routines and design the appropriate pendently calculated using Flexible-Meccano (Ozenne et al. 1 3 Journal of Biomolecular NMR (2018) 71:1–9 7 Fig. 4 A schematic representation of the Comparative/Stacking Anal- the z axis—and compare the derived results as a function of an exter- ysis workflow. Farseer-NMR can easily generate combined data from nal variable (e.g. ligand concentration, corresponding to the x axis of experiments (e.g. Intensity ratios for each residue between paramag- the Farseer-NMR’s Cube) as a single combined plot or as a PRE ver- netic and diamagnetic samples in PRE experiments—generated along sus ligand concentration curves for individual residues 2012) but incorporated into Farseer-NMR as additional (syn- error-prone manual analysis to a few minutes. Farseer- thetic) data. Farseer-NMR generated plots showed a direct NMR closes a longstanding gap in the quest for creating comparison of predicted and measured PRE, smoothed dif- an automated BioNMR pipeline. It is important to con- ference prol fi es obtained using a running Gaussian convolu - sider also that BioNMR-related peaklists are often ana- tion filter, heat-map representation of the ΔPRE data, side by lysed iteratively, that is, after restraint calculation and side comparisons, and per-residue chemical shift perturba- representation, further inspection of the spectra is often tion mapping of the whole constructs. This analysis clearly necessary to correct for interpretational errors, which identified departures from the random coil model and the may only be evident after considerable analysis. Farseer- role of the conserved aromatic residues in the formation of NMR drastically reduces the time required in this pro- a fuzzy complex in the N-terminal region of the Src protein. cess by enabling iterative analysis of a complete dataset Technically, analysis of paramagnetic NMR data consists with a single mouse “click”, using previously saved set- in representing the evolution of a paramagnetically gener- tings, which can be reloaded a posteriori. Farseer-NMR ated NMR parameter, in this case the ΔPRE calculated along was developed focusing on a “box”-like concept, that is, the z axis, along another axis that describes the system as input → routine → output, where routines can be as many a function of an external variable (sequence variants, point as imaginable. Rather than being a “black box”, Farseer- mutations, probe position). Farseer-NMR implements this NMR is an open-source “transparent box” where routines algorithm under the name “Comparative/Stacking Analysis”. can be inspected, improved and implemented by the users It serves as the basis for analysing paramagnetic NMR data themselves, enthusiast developers or via user-developer as well as a parsing routine to regroup results for improved interactions. In this vein, Farseer-NMR has been designed visualization. Figure  4 schematises the workflow of the to be user-friendly to NMR users with biological back- Comparative Analysis, described in detail in Sect. II-d of ground, but also as flexible as possible to engage advanced the Farseer-NMR Documentation. NMR users. The software is written in Python 3.6 related libraries and is stored in a GitHub repository, which is open-source and free-to-use according to the terms of the Concluding remarks GPL-3.0 license. We invite users to fork the repository and create pull requests to contribute to the Farseer-NMR Farseer-NMR is a software package to streamline and project. Indeed, we have developed Farseer-NMR with the standardize the analysis of NMR data obtained from scope of creating a nucleation point for specialized NMR sets of samples exploring multiple variations in terms of analysis routines that exist throughout the NMR commu- protein sequence or experimental conditions altering the nity so that those can become available to the community chemical nature of the sample, including ligand binding, as a whole in a easy and convenient manner. or its spectroscopic properties (e.g. paramagnetic and Acknowledgements This work was supported by the Spanish diamagnetic forms or isotropic versus oriented samples). MINECO (BIO2016-78006R), co-financed with EU structural funds Farseer-NMR facilitates the analysis of multivariable and the Fundació Marató TV3 (20132830/31) (JMCT, MA and MP), data by allowing the extraction of the individual contri- and by the Medical Research Council Project Grant MR/P000355/1 bution of the variables, and the identification and char - (SPS and ALB). We acknowledge access to the LRB NMR of the University of Barcelona. The authors are grateful to Susana Barrera- acterization of correlated effects and directly produces Vilarmau (ORCID 0000-0003-4868-6593) for all the interesting inputs publication-quality plots, reducing weeks of tedious and 1 3 8 Journal of Biomolecular NMR (2018) 71:1–9 during the development of this first release and to Jamie Ferrar (artis- Johnson BA, Blevins RA (1994) NMR view: a computer program ticsystems.uk) for providing the UI branding. for the visualization and analysis of NMR data. J Biomol NMR 4:603–614. https ://doi.org/10.1007/BF004 04272 Jung Y-S, Zweckstetter M (2004a) Backbone assignment of proteins Open Access This article is distributed under the terms of the Crea- with known structure using residual dipolar couplings. J Biomol tive Commons Attribution 4.0 International License (http://creat iveco NMR 30:25–35. https ://doi.org/10.1023/B:JNMR.00000 42955 mmons.or g/licenses/b y/4.0/), which permits unrestricted use, distribu- .14647 .77 tion, and reproduction in any medium, provided you give appropriate Jung YS, Zweckstetter M (2004b) Mars: robust automatic backbone credit to the original author(s) and the source, provide a link to the assignment of proteins. J Biomol NMR 30:11–23. https ://doi. Creative Commons license, and indicate if changes were made. org/10.1023/B:JNMR.00000 42954 .99056 .ad Kay LE (2016) New views of functionally dynamic proteins by solu- tion NMR spectroscopy. J Mol Biol 428:323–331. https ://doi. org/10.1016/j.jmb.2015.11.028 Keller R (2003) The CARA/Lua programmers manual. DOTANAL AG References Lee W, Tonelli M, Markley JL (2015) NMRFAM-SPARKY: enhanced software for biomolecular NMR spectroscopy. Bioinformatics Arai M, Ferreon JC, Wright PE (2012) Quantitative analysis of multi- 31:1325–1327. https ://doi.org/10.1093/bioin forma tics/btu83 0 site protein–ligand interactions by NMR: binding of intrinsically Maciejewski MW, Schuyler AD, Gryk MR, Moraru II, Romero PR, disordered p53 transactivation subdomains with the TAZ2 domain Ulrich EL, Eghbalnia HR, Livny M, Delaglio F, Hoch JC (2017) of CBP. J Am Chem Soc 134:3792–3803. https://doi.or g/10.1021/ NMRbox: a resource for biomolecular NMR computation. Bio- ja209 936u phys J 112:1529–1534. https:/ /doi.org/10.1016/j.bpj.2017.03.011 Arbesú M, Maffei M, Cordeiro TN, Teixeira JMC, Pérez Y, Bernadó P, Mahawaththa MC, Pearce BJG, Szabo M, Graham B, Klein CD, Roche S, Pons M (2017) The unique domain forms a fuzzy intra- Nitsche C, Otting G (2017) Solution conformations of a linked molecular complex in Src family kinases. Structure 25:630–640. construct of the Zika virus NS2B-NS3 protease. Antiviral Res e4. https ://doi.org/10.1016/j.str.2017.02.011 142:141–147. https ://doi.org/10.1016/j.antiv iral.2017.03.011 Bartels C, Xia TH, Billeter M, Güntert P, Wüthrich K (1995) The Mandel AM, Akke M, Palmer AG III (1995) Backbone dynamics of program XEASY for computer-supported NMR spectral analysis Escherichia coli ribonuclease HI: correlations with structure and of biological macromolecules. J Biomol NMR 6:1–10. https://doi. function in an active enzyme. J Mol Biol 246:144–163. https :// org/10.1007/BF004 17486 doi.org/10.1006/jmbi.1994.0073 Bijlmakers M-J, Teixeira JMC, Boer R, Mayzel M, Puig-Sàrries P, Marimon O, Teixeira JMC, Cordeiro TN, Soo VWC, Wood TL, Mayzel Karlsson G, Coll M, Pons M, Crosas B (2016) A C2HC zinc finger M, Amata I, García J, Morera A, Gay M, Vilaseca M, Orekhov is essential for the RING-E2 interaction of the ubiquitin ligase VY, Wood TK, Pons M (2016) An oxygen-sensitive toxin–anti- RNF125. Sci Rep 6:29232. https ://doi.org/10.1038/srep2 9232 toxin system. Nat Commun 7:13634. https ://doi.or g/10.1038/ Carlon A, Ravera E, Andrałojć W, Parigi G, Murshudov GN, Luchinat ncomm s1363 4 C (2016) How to tackle protein structural data from solution and McKinney W (2010) Data structures for statistical computing in solid state: an integrated approach. Prog Nucl Magn Reson Spec- python. In: Proceedings of the 9th Python in Science Conference, trosc 92–93:54–70. https ://doi.org/10.1016/j.pnmrs .2016.01.001 pp 51–56 d’Auvergne EJ, Gooley PR (2008) Optimisation of NMR dynamic mod- Mylona A, Theillet F-X, Foster C, Cheng TM, Miralles F, Bates PA, els I. Minimisation algorithms and their performance within the Selenko P, Treisman R (2016) Opposing effects of Elk-1 multisite model-free and Brownian rotational diffusion spaces. J Biomol phosphorylation shape its response to ERK activation. Science NMR 40:107–119. https ://doi.org/10.1007/s1085 8-007-9214-2 354:233–237. https ://doi.org/10.1126/scien ce.aad18 72 Delaglio F, Grzesiek S, Vuister GW, Zhu G, Pfeifer J, Bax A (1995) Narayanan RL, Dürr UHN, Bibow S, Biernat J, Mandelkow E, Zweck- NMRPipe: a multidimensional spectral processing system based stetter M (2010) Automatic assignment of the intrinsically disor- on UNIX pipes. J Biomol NMR 6:277–293 dered protein Tau with 441-residues. J Am Chem Soc 132:11906– Diniz A, Dias JS, Jiménez-Barbero J, Marcelo F, Cabrita EJ (2017) 11907. https ://doi.org/10.1021/ja105 657f Protein–glycan quinary interactions in crowding environment Orekhov VY, Nolde DE, Golovanov AP, Korzhnev DM, Arseniev AS unveiled by NMR spectroscopy. Chem - A Eur J 23:13213–13220. (1995) Processing of heteronuclear NMR relaxation data with the https ://doi.org/10.1002/chem.20170 2800 new software DASHA. Appl Magn Reson 9:581–588. https://doi. Güntert P (2009) Automated structure determination from NMR spec- org/10.1007/BF031 62365 tra. Eur Biophys J 38:129–143. https ://doi.or g/10.1007/s0024 Ozenne V, Bauer F, Salmon L, Huang J-R, Jensen MR, Segard S, Ber- 9-008-0367-z nadó P, Charavay C, Blackledge M (2012) Flexible-meccano: Güntert P, Mumenthaler C, Wüthrich K (1997) Torsion angle dynamics a tool for the generation of explicit ensemble descriptions of for NMR structure calculation with the new program DYANA. J intrinsically disordered proteins and their associated experi- Mol Biol 273:283–298. https ://doi.org/10.1006/jmbi.1997.1284 mental observables. Bioinformatics 28:1463–1470. https ://doi. Herrmann T, Güntert P, Wüthrich K (2002) Protein NMR structure org/10.1093/bioin forma tics/bts17 2 determination with automated NOE assignment using the new Pettersen EF, Goddard TD, Huang CC, Couch GS, Greenblatt DM, software CANDID and the torsion angle dynamics algorithm Meng EC, Ferrin TE (2004) UCSF Chimera: a visualization DYANA. J Mol Biol 319:209–227. https://doi.or g/10.1016/S0022 system for exploratory research and analysis. J Comput Chem -2836(02)00241 -3 25:1605–1612. https ://doi.org/10.1002/jcc.20084 Hunter JD (2007) Matplotlib: a 2D graphics environment. Comput Sci Pintacuda G, Kaikkonen A, Otting G (2004) Modulation of the dis- Eng 9:90–95. https ://doi.org/10.1109/MCSE.2007.55 tance dependence of paramagnetic relaxation enhancements by John M, Schmitz C, Park AY, Dixon NE, Huber T, Otting G (2007) CSA × DSA cross-correlation. J Magn Reson 171:233–243. https Sequence-specific and stereospecific assignment of methyl groups ://doi.org/10.1016/j.jmr.2004.08.019 using paramagnetic lanthanides. J Am Chem Soc 129:13749– Rieping W, Habeck M, Bardiaux B, Bernard A, Malliavin TE, Nilges M 13757. https ://doi.org/10.1021/ja074 4753 (2007) ARIA2: automated NOE assignment and data integration 1 3 Journal of Biomolecular NMR (2018) 71:1–9 9 in NMR structure calculation. Bioinformatics 23:381–382. https Theillet F-X, Smet-Nocca C, Liokatis S, Thongwichian R, Kosten ://doi.org/10.1093/bioin forma tics/btl58 9 J, Yoon M-K, Kriwacki RW, Landrieu I, Lippens G, Selenko P Schilder J, Liu W-M, Kumar P, Overhand M, Huber M, Ubbink M (2012) Cell signaling, post-translational protein modifications (2016) Protein docking using an ensemble of spin labels opti- and NMR spectroscopy. J Biomol NMR 54:217–236. https ://doi. mized by intra-molecular paramagnetic relaxation enhancement. org/10.1007/s1085 8-012-9674-x Phys Chem Chem Phys. https ://doi.org/10.1039/C5CP0 3781F Theillet F-X, Rose HM, Liokatis S, Binolfi A, Thongwichian R, Stuiver Schmitz C, John M, Park AY, Dixon NE, Otting G, Pintacuda G, Huber M, Selenko P (2013) Site-specific NMR mapping and time- T (2006) Efficient chi-tensor determination and NH assignment resolved monitoring of serine and threonine phosphorylation in of paramagnetic proteins. J Biomol NMR 35:79–87. https ://doi. reconstituted kinase reactions and mammalian cell extracts. Nat org/10.1007/s1085 8-006-9002-4 Protoc 8:1416–1432. https ://doi.org/10.1038/nprot .2013.083 Schmitz C, Stanton-Cook MJ, Su X-C, Otting G, Huber T (2008) van der Walt S, Colbert SC, Varoquaux G (2011) The NumPy array: Numbat: an interactive software tool for fitting Deltachi-tensors a structure for efficient numerical computation. Comput Sci Eng to molecular coordinates using pseudocontact shifts. J Biomol 13:22–30. https ://doi.org/10.1109/MCSE.2011.37 NMR 41:179–189. https ://doi.org/10.1007/s1085 8-008-9249-z Vranken WF, Boucher W, Stevens TJ, Fogh RH, Pajon A, Llinas M, Sengupta I, Bhate SH, Das R, Udgaonkar JB (2017) Salt-mediated Ulrich EL, Markley JL, Ionides J, Laue ED (2005) The CCPN oligomerization of the mouse prion protein monitored by real- data model for NMR spectroscopy: development of a software time NMR. J Mol Biol 429:1852–1872. https://doi.or g/10.1016/j. pipeline. Proteins Struct Funct Genet 59:687–696. h t tp s : / /d o i . jmb.2017.05.006 org/10.1002/prot.20449 Skinner SP, Moshev M, Hass MAS, Keizers PHJ, Ubbink M (2013) Wassenaar T, Dijk M, Loureiro-Ferreira N, Schot G, Vries SJ, Schmitz PARAssign—paramagnetic NMR assignments of protein nuclei C, Zwan J, Boelens R, Giachetti A, Ferella L, Rosato A, Bertini on the basis of pseudocontact shifts. J Biomol NMR 55:379–389. I, Herrmann T, Jonker HR, Bagaria A, Jaravine V, Güntert P, https ://doi.org/10.1007/s1085 8-013-9722-1 Schwalbe H, Vranken WF, Doreleijers JF, Vriend G, Vuister GW, Skinner SP, Fogh RH, Boucher W, Ragan TJ, Mureddu LG, Vuister Franke D, Kikhney A, Svergun DI, Fogh RH, Ionides J, Laue ED, GW (2016) CcpNmr AnalysisAssign: a flexible platform for inte- Spronk C, Jurkša S, Verlato M, Badoer S, Pra SD, Mazzucato M, grated NMR analysis. J Biomol NMR 66:111–124. https ://doi. Frizziero E, Bonvin AMJJ. (2012) WeNMR: structural biology org/10.1007/s1085 8-016-0060-y on the grid. J Grid Comput 10:743–767. https ://doi.org/10.1007/ Sormanni P, Piovesan D, Heller GT, Bonomi M, Kukic P, Camilloni s1072 3-012-9246-z C, Fuxreiter M, Dosztanyi Z, Pappu RV, Babu MM, Longhi S, Williamson MP (2013) Using chemical shift perturbation to charac- Tompa P, Dunker AK, Uversky VN, Tosatto SCE, Vendruscolo terise ligand binding. Prog Nucl Magn Reson Spectrosc 73:1–16 M (2017) Simultaneous quantification of protein order and disor - Zimmerman DE, Kulikowski CA, Huang Y, Feng W, Tashiro M, der. Nat Chem Biol 13:339–342. https ://doi.org/10.1038/nchem Shimotakahara S, Chien C, Powers R, Montelione GT (1997) bio.2331 Automated analysis of protein NMR assignments using methods Teilum K, Kunze MBA, Erlendsson S, Kragelund BB (2017) (S)Pin- from artificial intelligence. J Mol Biol 269:592–610. https ://doi. ning down protein interactions by NMR. Protein Sci 26:436–451. org/10.1006/jmbi.1997.1052 https ://doi.org/10.1002/pro.3105 1 3 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Biomolecular NMR Springer Journals

Farseer-NMR: automatic treatment, analysis and plotting of large, multi-variable NMR data

Free
9 pages
Loading next page...
 
/lp/springer_journal/farseer-nmr-automatic-treatment-analysis-and-plotting-of-large-multi-bRkTPpuiU5
Publisher
Springer Netherlands
Copyright
Copyright © 2018 by The Author(s)
Subject
Physics; Biological and Medical Physics, Biophysics; Biochemistry, general; Spectroscopy/Spectrometry
ISSN
0925-2738
eISSN
1573-5001
D.O.I.
10.1007/s10858-018-0182-5
Publisher site
See Article on Publisher Site

Abstract

We present Farseer-NMR (https://git.io/vA ueU), a sof tware package to treat, evaluate and combine NMR spectroscopic data from sets of protein-derived peaklists covering a range of experimental conditions. The combined advances in NMR and molecular biology enable the study of complex biomolecular systems such as flexible proteins or large multibody complexes, which display a strong and functionally relevant response to their environmental conditions, e.g. the presence of ligands, site- directed mutations, post translational modifications, molecular crowders or the chemical composition of the solution. These advances have created a growing need to analyse those systems’ responses to multiple variables. The combined analysis of NMR peaklists from large and multivariable datasets has become a new bottleneck in the NMR analysis pipeline, whereby information-rich NMR-derived parameters have to be manually generated, which can be tedious, repetitive and prone to human error, or even unfeasible for very large datasets. There is a persistent gap in the development and distribution of soft- ware focused on peaklist treatment, analysis and representation, and specifically able to handle large multivariable datasets, which are becoming more commonplace. In this regard, Farseer-NMR aims to close this longstanding gap in the automated NMR user pipeline and, altogether, reduce the time burden of analysis of large sets of peaklists from days/weeks to seconds/ minutes. We have implemented some of the most common, as well as new, routines for calculation of NMR parameters and several publication-quality plotting templates to improve NMR data representation. Farseer-NMR has been written entirely in Python and its modular code base enables facile extension. Keywords NMR spectroscopy · Data analysis · Intrinsically disordered proteins · Paramagnetic-NMR · Proteins · Chemical shift perturbations Introduction Electronic supplementary material The online version of this article (https ://doi.org/10.1007/s1085 8-018-0182-5) contains Flexible proteins, including fully disordered proteins, short supplementary material, which is available to authorized users. disordered segments or interdomain linkers, as well as multi- body macromolecular complexes, play a fundamental role * João M. C. Teixeira in many key regulatory processes by which environmental joaomcteixeira@gmail.com information is sensed, processed or used to modulate com- Simon P. Skinner plex responses. The reaction of these systems to environ- s.p.skinner@leeds.ac.uk mental changes, which provide fundamental information on Miguel Arbesú their function, can be experimentally investigated by explor- miguelarbesu@ub.edu ing them under a large set of conditions and evaluating their Alexander L. Breeze multivariable response. In this regard, Nuclear magnetic res- a.l.breeze@leeds.ac.uk onance (NMR) generates atomic level data bearing structural Miquel Pons and dynamic information. Over the last two decades, the mpons@ub.edu information that can be extracted from NMR observables BioNMR Group, Inorganic and Organic Chemistry has continuously increased owing to the development of Department, University of Barcelona, Barcelona, Spain sophisticated NMR methods combined with molecular biol- Astbury Centre for Structural Molecular Biology, Faculty ogy protocols for sample preparation, which together have of Biological Sciences, University of Leeds, Leeds, UK Vol.:(0123456789) 1 3 2 Journal of Biomolecular NMR (2018) 71:1–9 considerably enriched the NMR spectroscopist’s toolkit. The NMR-peaklist text files which contain the NMR observa- advent of these advances places NMR in the front line as bles. Moreover, there are multiple ways of combining the one of the most valuable techniques to investigate complex variable-dependent information, increasing the complex- protein systems under a multitude of conditions (Kay 2016; ity of the analysis and leading to an overload of repetitive Sormanni et al. 2017), which include the mapping of ligand tasks if peaklists are treated manually. binding sites and their affinities (Arai et al. 2012; Teilum In addition to the above arguments, the nature of NMR et al. 2017), the presence of molecular crowders (Diniz et al. users, and consequently NMR software users, has also 2017), different chemical compositions of the solution (Sen- evolved over the past two decades, from specialist NMR gupta et al. 2017) as well as site-directed mutations (Arbesú spectroscopists to structural biologists, who are not nec- et al. 2017) and local post-translational modifications (Theil- essarily NMR experts, but use the technique routinely for let et al. 2012) whose effects expand to the whole molecule. data acquisition and analysis. As a result of the evolution Such experimental design generates large and multivariable of the biomolecular NMR user base, there is an increasing sets of NMR data. Complex datasets are also generated in need for the development of a user-directed NMR pipeline, the context of time-resolved NMR studies (Theillet et al. which focuses on the needs of the community as a whole. 2013; Mylona et al. 2016), and structural/dynamic analy- The interplay between NMR and other techniques for struc- sis by paramagnetic observables (Mahawaththa et al. 2017) tural biology is accelerating this paradigm shift (Wassenaar from different sources (Carlon et al. 2016; Schilder et al. et al. 2012, http://www.inext -eu.org). Moreover, we believe 2016). Nonetheless, exploiting a combined use of such avail- that such a pipeline should be open, easily extensible and able NMR tools is not yet routine, mostly due to a lack of accessible to anyone who wants to use and contribute to it, platforms that close the gap between the expertise of devel- in compliance with the most up-to-date Open Access poli- opers and the users’ needs, both at the acquisition and the cies of the European Community (http://ec.europ a.eu/resea analysis level.rch/openscienc e/inde x.cfm ), and not closed to all except the Numerous NMR data analysis packages have been software’s developers. developed over the past 25 years (Maciejewski et al. 2017); Taken together, the above arguments have motivated us this includes packages for specific applications e.g. relaxa- to develop Farseer-NMR, a user-directed, open source and tion analysis (Mandel et al. 1995; Orekhov et al. 1995; modular toolbox into which users can input and analyse d’Auvergne and Gooley 2008), structure solution (Güntert large datasets of multivariable NMR curated peaklists in an et  al. 1997; Herrmann et al. 2002; Rieping et al. 2007; efficient, reproducible and organized manner and without Güntert 2009), paramagnetic NMR analysis (Pintacuda a steep learning curve. The current version has been spe- et al. 2004; Schmitz et al. 2006, 2008; John et al. 2007; cifically designed for the analysis of protein-related data. Skinner et al. 2013), resonance assignment (Zimmerman Farseer-NMR transforms the analysis of numerous peaklists et al. 1997; Jung and Zweckstetter 2004a, b; Narayanan from weeks of work to a few minutes of automatic execution, et  al. 2010), and comprehensive packages to perform a as is represented in Fig. 1 of Farseer-NMR Documentation multitude of different analyses mainly focused on spectral provided in Supplementary Material, something which has processing and data representation, whilst also taking care been lacking in the NMR analysis pipeline. The software is of the associated bookkeeping (Johnson and Blevins 1994; coded using the Python programming language. We pro- Bartels et  al. 1995; Delaglio et  al. 1995; Keller 2003; pose Farseer-NMR also as a nucleation point for the vari- Vranken et al. 2005; Lee et al. 2015; Skinner et al. 2016). ous NMR software packages/routines built to treat NMR- In spite of the plethora of analysis that these packages can derived parameters, in particular those without a graphical perform, which unarguably constitute milestone achieve- user interface (GUI). We have endeavoured to facilitate this ments in the field, we persistently find a gap on software by providing a platform upon which additional routines can availability dedicated to conversion of NMR observables be developed and launched. Farseer-NMR has been devel- in large sets of peaklists into correlated information-rich oped around the concept of a “transparent box”, rather than a parameters from which biologically relevant answers can “black box”, or just another Python library for programmers be extracted and comfortably presented to the community. to use. In this respect, we distribute Farseer-NMR with a This gap expands as the complexity of the systems under complete documentation record which, together with inline the scope of biomolecular NMR projects increases and code commentaries, explain the program structure and archi- the number of variables needed to characterise them also tecture, provide examples of new implementations, both grows. Hence, it is often necessary to acquire a multitude for core and for the user interface, and describe the setup of NMR spectra covering this whole range of variables— and execution of a Farseer-NMR calculation. All settings we term these datasets multivariable. Upon spectral analy- required for a Farseer-NMR calculation are user-settable, sis, the researcher faces the task of extracting structural either using a configuration file, or via the GUI. This report information from correlated series of, maybe hundreds, describes the concepts of Farseer-NMR, its data structure 1 3 Journal of Biomolecular NMR (2018) 71:1–9 3 alpha-release, it is fully operational, but very much open to improvements that will be directed by the interaction with the community. The artwork in this manuscript was prepared using the Matplotlib functions developed for Farseer-NMR, the LibreOffice Suite (https ://www.libre offic e.org/), Adobe Photoshop CS2 (http://www .adobe .com ) and assembled with GIMP (https ://www.gimp.org/). The Farseer‑NMR structure Farseer-NMR has been developed to enable analysis of mul- tivariable sets of protein-related NMR data in a concerted, correlated and automated manner; variables can be continu- ous e.g. concentration of a titrated ligand, or discreet e.g. dif- ferent protein constructs, ligands, environment. User curated peaklists from two-dimensional NMR spectra are imported Fig. 1 A schematic representation of the Farseer-NMR Cube. Exam- ples of different variables are given of each Cube’s axis. Each col- into the Farseer-NMR data structure, and from here, experi- oured rectangle, the first one highlighted with ‘.csv’ for representa- mental datasets can be freely navigated and exploited for tion is a 2D-NMR peaklist analysis without any restrictions of the acquisition schedule. Up to three experimental variables can be simultaneously and implementation of its functionalities, all of which do analysed by Farseer-NMR and, therefore, we refer to the not require any programming knowledge. data structure as the “Farseer-NMR cube” (Fig. 1 and Sect. II-a of the Documentation PDF), on which each dimension/ axis, x, y or z, of the cube represents a different experimen- Methods tal variable and each cube data point, a coordinate within these axes, is an NMR peaklist table composed of rows that The Farseer-NMR Project is hosted at GitHub (https://git hu represent the protein residues and columns that contain all b.com) and distributed under the GPL-3.0 license (https :// residue-related information extracted from the NMR spectra www.fsf.org/); the following link directs to the project front alongside user annotations. This configuration yields a final page: https://git.io/vA ueU. F arseer-NMR was written using 5D matrix-like structure stored in a single digital object. version 3.6 of the Python programming language, and vari- The Farseer-NMR Cube can be traversed along the different ous Python libraries were used for data management and dimensions describing the dependency of the data on the analysis. Peaklist treatment and management, parameter different defined variables. calculation and data fitting are performed using a combi - For example, if a data set contains diamagnetic and para- nation of routines from Numpy and Scipy (van der Walt magnetic spectra of two different protein constructs, into et al. 2011) and Pandas (Wes 2010), and plotting templates which a ligand concentration range was titrated, Farseer- were developed using Matplotlib (Hunter 2007). The GUI NMR can analyse the spectral variations (for example, chem- was developed using PyQt v5.7. Version control of the code ical shifts, intensities, linewidths, couplings) with respect base was achieved using a git repository hosted by GitHub to ligand concentration and for each construct separately (http://www.github.com ). Farseer-NMR was developed and (extracting information that can be used to calculate binding tested using the GNU/Linux operating system and Anaconda affinities, x axis), calculate the changes caused by a given Python (http://www.anaco nda.com) distribution, addition- ligand concentration between the two constructs (y axis), ally we provide Anaconda environment files for users to and extract paramagnetic relaxation enhancements (PRE) setup a running environment if necessary. We have also for all spectra by comparing the paramagnetic series with produced complete documentation containing a plethora the diamagnetic one (z axis), from which distance dependent of technical information about Farseer-NMR, its concepts information can be extracted. Additionally, data calculated and its functionalities. A maintained version of the Farseer- along different axes can be parsed and represented together NMR Documentation is available online from the home page along the other axes (comparative analysis, see Case Study). repository, the version corresponding to the current release The selection of different dimensions of the cube for analysis has been supplied as Supplementary Material. At the time and comparison across dimensions are Boolean flags in the of this manuscript, the version of Farseer-NMR is that of an setup of a Farseer-NMR calculation (vide infra). Axes can 1 3 4 Journal of Biomolecular NMR (2018) 71:1–9 have any number of data points (peaklists) without compro- Farseer‑NMR analysis routines mising the organization and representation of results. Any type of data can be specified along any of the differ - Each series extracted from the Farseer-NMR cube can be ent x, y, and z axes; however, for the sake of simplicity and analysed using a number of analysis routines, which are clarity, some analysis routines are currently restricted to spe- user-configurable. Some commonly used calculation and cific axes, e.g. paramagnetic analysis can only be performed plotting routines for NMR analysis of macromolecular data along the z axis, inclusion of different protein sequences can have been implemented in the Farseer-NMR code base; only be performed along the y axis, and fitting of continu - we have also included novel methods/templates among ous data can only be performed along the x axis. Complete which is the recently developed ΔPRE method (Arbesú descriptions of axis characteristics can be found in Sect. V-d et al. 2017). In addition, and more importantly, the core of the Farseer-NMR Documentation. of Farseer-NMR has been designed such that virtually any calculation or plotting routine can be implemented. Far- seer-NMR analysis routines consist of three steps, namely, The Farseer‑NMR workflow (i) calculation of NMR parameters from raw data, (ii) data and values export and (iii) plotting. The current version One of the great advantages of having all experimental data of Farseer-NMR can calculate nucleus-specific and com- in one computational object, the Farseer-NMR Cube, is the bined chemical shift perturbations (CSPs), according to ability to slice the cube in any direction and interrogate the Williamson (2013), height and volume ratios, and ΔPRE data to investigate specific questions or phenomena which values according to Arbesú et al. (2017). Elaborated NMR are not limited to the acquisition schedule. The first step parameters are calculated by evaluating each peaklist in an of any Farseer-NMR calculation is to load the user curated experimental series with the appropriate reference experi- peaklists as digital tables. Farseer-NMR accepts as input ment, which is the first in the series, and the calculated peaklists from a variety of formats, namely, Ansig, CCPN values are stored in the corresponding peaklist in newly Analysis version 2 (Vranken et al. 2005), NMRPipe (Dela- added columns. Farseer-NMR also contains a restraint fit- glio et al. 1995), NmrView (Johnson and Blevins 1994) and ting platform, whereby continuous data can be fit, cur - Sparky (Lee et al. 2015), taking a step towards the integra- rently restricted to data contained along the x axis. It is not tion of the NMR community as a whole. Once imported, the aim of Farseer-NMR to provide, presently, a complete the peaklists are converted into an enhanced format initially suite of routines for fitting the most diverse natures of derived from CCPN Analysis version 2 and are stored in a continuous NMR data, instead, to provide a platform for variable-defined hierarchical directory structure created by which users/developers can implement their own fitting Farseer-NMR. The peaklists dataset is loaded to memory as routines that can plug into the core Farseer-NMR routines, a nested dictionary. At this stage, peaklists are scanned for and, in this way, open the door for an easy distribution identification of missing and/or unassigned residues along among the whole NMR community. the first variable (x axis) and new rows representing these During the calculation run, a set of folders, organised residues are added to the peaklists. This serves two pur- hierarchically according to the defined variables, is cre- poses: identification of those residues and normalization of ated to store all Farseer-NMR output, so that the user can the peaklists to the same number of rows; these procedures readily access the generated results. Within these folders, are described in detail in Sect. II-c of the Farseer-NMR formatted, treated and parsed peaklists are exported as Documentation. Amide sidechain entries are also identified comma-separated files, which contain the newly calcu- at this stage, if required. Following peaklist treatment, the lated NMR parameters in addition to the original data. Farseer-NMR Cube is generated. Moreover, the user-specified plots are created alongside All settings required by Farseer-NMR for a calculation: parsed text tables containing only the represented data. selection of dimensions for analysis, types of analysis, types UCSF Chimera (Pettersen et al. 2004) attribute files are of output plots and settings for the individual plot types, generated to allow straightforward representation of the are specified in a JSON file (http://www.json.org/), which calculated restraints in 3D molecular structures; facile can be prepared manually or via the GUI (vide infra). Once extension of this feature to meet other molecular repre- executed, Farseer-NMR sequentially generates series of sentation software requirements is possible. At present, peaklists from the Farseer-NMR cube, which represent Farseer-NMR contains eight plotting templates (three bar the system’s evolution along an axis/variable at fixed data plots, two scatter plots, a heat map and two continuous points of the other two axes and walks through all the pos- data plots), which have been prepared specifically for sible combinations of x, y and z variables, (Sect. II-b of NMR-derived data representation. These represent some the Farseer-NMR Documentation) and analyses these series of the most common plotting styles in NMR literature, according to the user-specified settings. 1 3 Journal of Biomolecular NMR (2018) 71:1–9 5 along with novel methods of presenting NMR-derived The graphical user interface (GUI) data. The plots generated by Farseer-NMR are composed of subplots that either represent data for all residues in To facilitate easy preparation and execution of Farseer-NMR a given series, or residue-specific variations for a given calculation runs, a user-friendly GUI has been developed; variable (where appropriate). This enables facile identi- full description can be found in Sect. III of the Farseer-NMR fication of significant results from a single figure, either Documentation. The current GUI is divided into two tabs, by condition or by residue. Default configurations for all namely, Peaklist Selection (Fig. 3) and Settings (Sect. III-c plots have been implemented, which yield publication- of the Documentation PDF). In similarity with the core code, quality figures, although templates can be fully config- the GUI was developed to be easily expansible and designed ured by the user. Figure  2 shows examples of three of so that new features can be added as new tabs with their the plotting templates while full information regarding respective options and functions. We envisage user-devel- Farseer-NMR plotting features is available in Sect. VI-d oper contributions playing a significant role in this evolution of the Farseer-NMR Documentation. If continuous data of the interface. The Peaklist Selection tab consists of three fitting is performed as part of a Farseer-NMR calculation distinct areas: a sidebar for data import and selection, an run, the values obtained from fitting can be represented as area for specifying experimental conditions, and an area for a plot of restraint evolution per residue. Farseer-NMR has construction of the Farseer-NMR data structure. Peaklists already been used to generate figures in published work are imported by dropping them on to the side bar, which will (Bijlmakers et  al. 2016; Marimon et  al. 2016; Arbesú detect the peaklist format and import the data if the peaklist et al. 2017). format is recognised (vide supra). Setting up the Farseer- NMR data structure consists of two stages: (1) specifying the numbers of points and labels for each of the axes of the Far- seer-NMR cube, that is, experimental variables and number Fig. 2 Three examples of the implemented Farseer-NMR plotting lines are identified by character “P”. Blue and gold bars (labelled 1 1 15 templates. H and N chemical shifts were generated from a syn- and 2, respectively) represent two pairs of residues with alternative thetic data set simulating seven points of a ligand titration of a 100 assignment, that after plot analysis can be swapped with confidence residues protein. a Extended bar plot template and b compacted bar (14F ↔ 48R, 18Q ↔ 97M), colours and labels are representations plot template representing the combined chemical shift perturbations of the user annotations present in the original peaklists. A signifi- (CSP), calculated according to Williamson (2013), for the last experi- cance threshold is represented as a red line. c Progression of the CSP ment in the series (with 800 µM ligand) versus the reference experi- parameter represented individually for each residue. Peaks that have ment (ligand free). Black bars represent residues measured in that disappeared along the series are easily identified (22F), as well as spectrum. Unassigned residues are represented in grey a × ticks or b unassigned peaks. The represented plots are crops of the full pictures background shade. Red bars represent missing residues, i.e. residues generated by Farseer-NMR that represent the whole series. All col- that have been observed previously in the series but have disappeared ours, sizes, shapes, and labels are user-configurable at a given ligand concentration; the last measured value is kept. Pro- 1 3 6 Journal of Biomolecular NMR (2018) 71:1–9 Fig. 3 A screenshot of the Farseer-NMR user interface. The Peaklist selection tab is shown loaded with an artificial dataset as an example. The Farseer-NMR Cube variables are set up and the corresponding data points are populated with peaklists of measured data points, respectively, and (2) selection of setting menus under a new tab. Default settings for all of the appropriate peaklist for each combination of variables. the fields, are set at start-up, and configuration files can also Once the numbers of points, and labels for each axis point be loaded and saved from this tab. The footer of the GUI have been specified, a hierarchical tree diagram representing enables instant access to Farseer-NMR documentation, to all specified conditions is drawn with “Drop peaklist here”, the Farseer-NMR Twitter account, and to contact the devel- written at the end of each branch. From here, peaklists cor- opers via e-mail to submit feedback, request features and responding to the correct combination of variables can be report any bugs. dragged and dropped onto the end of each branch of the tree, thereby creating the Farseer-NMR data structure. The Settings tab of the GUI contains input fields for all Case study: ΔPRE analysis the parameters needed to analyse the data using Farseer- NMR. There are two text input boxes, one for setting An example of the versatility of Farseer-NMR is the imple- the Peaklist Dataset Folder from which peaklists can be mentation of a new procedure for the analysis of intramo- imported aside from the drag-and-drop area, and a second lecular PRE induced by a paramagnetic probe on disordered for setting the Calculation Output Path to which all treated protein fragments of a series of proteins harboring mutations peaklists and plots will be written. Selection of axes along that may affect their dynamics and conformational properties which to perform analysis, and options to compare across (Arbesú et al 2017). The data set was formed of 10 sequence dimensions, consider side chain amide peaks and identifi- variants that probed the effect of four point mutations at cation of missing and unassigned residues, are specified by conserved aromatic residues and the presence of the folded checkboxes in this tab. Settings for chemical shift normalisa- domain adjacent to the disordered region by measuring the tion, NMR observable selection and parameter calculation, relaxation induced by paramagnetic centre incorporated in plotting template selection, figure format and file dimension three alternative sites along the sequence. Experimental data and resolution are also configurable from this tab. A series consisted of HSQC spectra of the paramagnetic proteins and of popups accessible from this tab contain all the settings the corresponding diamagnetic control (20 experiments). for all the implemented plotting templates. The GUI has The observed PREs were compared with the values expected been forged to facilitate, and invite, advanced users to imple- from a random coil model for the disordered region, inde- ment new calculation routines and design the appropriate pendently calculated using Flexible-Meccano (Ozenne et al. 1 3 Journal of Biomolecular NMR (2018) 71:1–9 7 Fig. 4 A schematic representation of the Comparative/Stacking Anal- the z axis—and compare the derived results as a function of an exter- ysis workflow. Farseer-NMR can easily generate combined data from nal variable (e.g. ligand concentration, corresponding to the x axis of experiments (e.g. Intensity ratios for each residue between paramag- the Farseer-NMR’s Cube) as a single combined plot or as a PRE ver- netic and diamagnetic samples in PRE experiments—generated along sus ligand concentration curves for individual residues 2012) but incorporated into Farseer-NMR as additional (syn- error-prone manual analysis to a few minutes. Farseer- thetic) data. Farseer-NMR generated plots showed a direct NMR closes a longstanding gap in the quest for creating comparison of predicted and measured PRE, smoothed dif- an automated BioNMR pipeline. It is important to con- ference prol fi es obtained using a running Gaussian convolu - sider also that BioNMR-related peaklists are often ana- tion filter, heat-map representation of the ΔPRE data, side by lysed iteratively, that is, after restraint calculation and side comparisons, and per-residue chemical shift perturba- representation, further inspection of the spectra is often tion mapping of the whole constructs. This analysis clearly necessary to correct for interpretational errors, which identified departures from the random coil model and the may only be evident after considerable analysis. Farseer- role of the conserved aromatic residues in the formation of NMR drastically reduces the time required in this pro- a fuzzy complex in the N-terminal region of the Src protein. cess by enabling iterative analysis of a complete dataset Technically, analysis of paramagnetic NMR data consists with a single mouse “click”, using previously saved set- in representing the evolution of a paramagnetically gener- tings, which can be reloaded a posteriori. Farseer-NMR ated NMR parameter, in this case the ΔPRE calculated along was developed focusing on a “box”-like concept, that is, the z axis, along another axis that describes the system as input → routine → output, where routines can be as many a function of an external variable (sequence variants, point as imaginable. Rather than being a “black box”, Farseer- mutations, probe position). Farseer-NMR implements this NMR is an open-source “transparent box” where routines algorithm under the name “Comparative/Stacking Analysis”. can be inspected, improved and implemented by the users It serves as the basis for analysing paramagnetic NMR data themselves, enthusiast developers or via user-developer as well as a parsing routine to regroup results for improved interactions. In this vein, Farseer-NMR has been designed visualization. Figure  4 schematises the workflow of the to be user-friendly to NMR users with biological back- Comparative Analysis, described in detail in Sect. II-d of ground, but also as flexible as possible to engage advanced the Farseer-NMR Documentation. NMR users. The software is written in Python 3.6 related libraries and is stored in a GitHub repository, which is open-source and free-to-use according to the terms of the Concluding remarks GPL-3.0 license. We invite users to fork the repository and create pull requests to contribute to the Farseer-NMR Farseer-NMR is a software package to streamline and project. Indeed, we have developed Farseer-NMR with the standardize the analysis of NMR data obtained from scope of creating a nucleation point for specialized NMR sets of samples exploring multiple variations in terms of analysis routines that exist throughout the NMR commu- protein sequence or experimental conditions altering the nity so that those can become available to the community chemical nature of the sample, including ligand binding, as a whole in a easy and convenient manner. or its spectroscopic properties (e.g. paramagnetic and Acknowledgements This work was supported by the Spanish diamagnetic forms or isotropic versus oriented samples). MINECO (BIO2016-78006R), co-financed with EU structural funds Farseer-NMR facilitates the analysis of multivariable and the Fundació Marató TV3 (20132830/31) (JMCT, MA and MP), data by allowing the extraction of the individual contri- and by the Medical Research Council Project Grant MR/P000355/1 bution of the variables, and the identification and char - (SPS and ALB). We acknowledge access to the LRB NMR of the University of Barcelona. The authors are grateful to Susana Barrera- acterization of correlated effects and directly produces Vilarmau (ORCID 0000-0003-4868-6593) for all the interesting inputs publication-quality plots, reducing weeks of tedious and 1 3 8 Journal of Biomolecular NMR (2018) 71:1–9 during the development of this first release and to Jamie Ferrar (artis- Johnson BA, Blevins RA (1994) NMR view: a computer program ticsystems.uk) for providing the UI branding. for the visualization and analysis of NMR data. J Biomol NMR 4:603–614. https ://doi.org/10.1007/BF004 04272 Jung Y-S, Zweckstetter M (2004a) Backbone assignment of proteins Open Access This article is distributed under the terms of the Crea- with known structure using residual dipolar couplings. J Biomol tive Commons Attribution 4.0 International License (http://creat iveco NMR 30:25–35. https ://doi.org/10.1023/B:JNMR.00000 42955 mmons.or g/licenses/b y/4.0/), which permits unrestricted use, distribu- .14647 .77 tion, and reproduction in any medium, provided you give appropriate Jung YS, Zweckstetter M (2004b) Mars: robust automatic backbone credit to the original author(s) and the source, provide a link to the assignment of proteins. J Biomol NMR 30:11–23. https ://doi. Creative Commons license, and indicate if changes were made. org/10.1023/B:JNMR.00000 42954 .99056 .ad Kay LE (2016) New views of functionally dynamic proteins by solu- tion NMR spectroscopy. J Mol Biol 428:323–331. https ://doi. org/10.1016/j.jmb.2015.11.028 Keller R (2003) The CARA/Lua programmers manual. DOTANAL AG References Lee W, Tonelli M, Markley JL (2015) NMRFAM-SPARKY: enhanced software for biomolecular NMR spectroscopy. Bioinformatics Arai M, Ferreon JC, Wright PE (2012) Quantitative analysis of multi- 31:1325–1327. https ://doi.org/10.1093/bioin forma tics/btu83 0 site protein–ligand interactions by NMR: binding of intrinsically Maciejewski MW, Schuyler AD, Gryk MR, Moraru II, Romero PR, disordered p53 transactivation subdomains with the TAZ2 domain Ulrich EL, Eghbalnia HR, Livny M, Delaglio F, Hoch JC (2017) of CBP. J Am Chem Soc 134:3792–3803. https://doi.or g/10.1021/ NMRbox: a resource for biomolecular NMR computation. Bio- ja209 936u phys J 112:1529–1534. https:/ /doi.org/10.1016/j.bpj.2017.03.011 Arbesú M, Maffei M, Cordeiro TN, Teixeira JMC, Pérez Y, Bernadó P, Mahawaththa MC, Pearce BJG, Szabo M, Graham B, Klein CD, Roche S, Pons M (2017) The unique domain forms a fuzzy intra- Nitsche C, Otting G (2017) Solution conformations of a linked molecular complex in Src family kinases. Structure 25:630–640. construct of the Zika virus NS2B-NS3 protease. Antiviral Res e4. https ://doi.org/10.1016/j.str.2017.02.011 142:141–147. https ://doi.org/10.1016/j.antiv iral.2017.03.011 Bartels C, Xia TH, Billeter M, Güntert P, Wüthrich K (1995) The Mandel AM, Akke M, Palmer AG III (1995) Backbone dynamics of program XEASY for computer-supported NMR spectral analysis Escherichia coli ribonuclease HI: correlations with structure and of biological macromolecules. J Biomol NMR 6:1–10. https://doi. function in an active enzyme. J Mol Biol 246:144–163. https :// org/10.1007/BF004 17486 doi.org/10.1006/jmbi.1994.0073 Bijlmakers M-J, Teixeira JMC, Boer R, Mayzel M, Puig-Sàrries P, Marimon O, Teixeira JMC, Cordeiro TN, Soo VWC, Wood TL, Mayzel Karlsson G, Coll M, Pons M, Crosas B (2016) A C2HC zinc finger M, Amata I, García J, Morera A, Gay M, Vilaseca M, Orekhov is essential for the RING-E2 interaction of the ubiquitin ligase VY, Wood TK, Pons M (2016) An oxygen-sensitive toxin–anti- RNF125. Sci Rep 6:29232. https ://doi.org/10.1038/srep2 9232 toxin system. Nat Commun 7:13634. https ://doi.or g/10.1038/ Carlon A, Ravera E, Andrałojć W, Parigi G, Murshudov GN, Luchinat ncomm s1363 4 C (2016) How to tackle protein structural data from solution and McKinney W (2010) Data structures for statistical computing in solid state: an integrated approach. Prog Nucl Magn Reson Spec- python. In: Proceedings of the 9th Python in Science Conference, trosc 92–93:54–70. https ://doi.org/10.1016/j.pnmrs .2016.01.001 pp 51–56 d’Auvergne EJ, Gooley PR (2008) Optimisation of NMR dynamic mod- Mylona A, Theillet F-X, Foster C, Cheng TM, Miralles F, Bates PA, els I. Minimisation algorithms and their performance within the Selenko P, Treisman R (2016) Opposing effects of Elk-1 multisite model-free and Brownian rotational diffusion spaces. J Biomol phosphorylation shape its response to ERK activation. Science NMR 40:107–119. https ://doi.org/10.1007/s1085 8-007-9214-2 354:233–237. https ://doi.org/10.1126/scien ce.aad18 72 Delaglio F, Grzesiek S, Vuister GW, Zhu G, Pfeifer J, Bax A (1995) Narayanan RL, Dürr UHN, Bibow S, Biernat J, Mandelkow E, Zweck- NMRPipe: a multidimensional spectral processing system based stetter M (2010) Automatic assignment of the intrinsically disor- on UNIX pipes. J Biomol NMR 6:277–293 dered protein Tau with 441-residues. J Am Chem Soc 132:11906– Diniz A, Dias JS, Jiménez-Barbero J, Marcelo F, Cabrita EJ (2017) 11907. https ://doi.org/10.1021/ja105 657f Protein–glycan quinary interactions in crowding environment Orekhov VY, Nolde DE, Golovanov AP, Korzhnev DM, Arseniev AS unveiled by NMR spectroscopy. Chem - A Eur J 23:13213–13220. (1995) Processing of heteronuclear NMR relaxation data with the https ://doi.org/10.1002/chem.20170 2800 new software DASHA. Appl Magn Reson 9:581–588. https://doi. Güntert P (2009) Automated structure determination from NMR spec- org/10.1007/BF031 62365 tra. Eur Biophys J 38:129–143. https ://doi.or g/10.1007/s0024 Ozenne V, Bauer F, Salmon L, Huang J-R, Jensen MR, Segard S, Ber- 9-008-0367-z nadó P, Charavay C, Blackledge M (2012) Flexible-meccano: Güntert P, Mumenthaler C, Wüthrich K (1997) Torsion angle dynamics a tool for the generation of explicit ensemble descriptions of for NMR structure calculation with the new program DYANA. J intrinsically disordered proteins and their associated experi- Mol Biol 273:283–298. https ://doi.org/10.1006/jmbi.1997.1284 mental observables. Bioinformatics 28:1463–1470. https ://doi. Herrmann T, Güntert P, Wüthrich K (2002) Protein NMR structure org/10.1093/bioin forma tics/bts17 2 determination with automated NOE assignment using the new Pettersen EF, Goddard TD, Huang CC, Couch GS, Greenblatt DM, software CANDID and the torsion angle dynamics algorithm Meng EC, Ferrin TE (2004) UCSF Chimera: a visualization DYANA. J Mol Biol 319:209–227. https://doi.or g/10.1016/S0022 system for exploratory research and analysis. J Comput Chem -2836(02)00241 -3 25:1605–1612. https ://doi.org/10.1002/jcc.20084 Hunter JD (2007) Matplotlib: a 2D graphics environment. Comput Sci Pintacuda G, Kaikkonen A, Otting G (2004) Modulation of the dis- Eng 9:90–95. https ://doi.org/10.1109/MCSE.2007.55 tance dependence of paramagnetic relaxation enhancements by John M, Schmitz C, Park AY, Dixon NE, Huber T, Otting G (2007) CSA × DSA cross-correlation. J Magn Reson 171:233–243. https Sequence-specific and stereospecific assignment of methyl groups ://doi.org/10.1016/j.jmr.2004.08.019 using paramagnetic lanthanides. J Am Chem Soc 129:13749– Rieping W, Habeck M, Bardiaux B, Bernard A, Malliavin TE, Nilges M 13757. https ://doi.org/10.1021/ja074 4753 (2007) ARIA2: automated NOE assignment and data integration 1 3 Journal of Biomolecular NMR (2018) 71:1–9 9 in NMR structure calculation. Bioinformatics 23:381–382. https Theillet F-X, Smet-Nocca C, Liokatis S, Thongwichian R, Kosten ://doi.org/10.1093/bioin forma tics/btl58 9 J, Yoon M-K, Kriwacki RW, Landrieu I, Lippens G, Selenko P Schilder J, Liu W-M, Kumar P, Overhand M, Huber M, Ubbink M (2012) Cell signaling, post-translational protein modifications (2016) Protein docking using an ensemble of spin labels opti- and NMR spectroscopy. J Biomol NMR 54:217–236. https ://doi. mized by intra-molecular paramagnetic relaxation enhancement. org/10.1007/s1085 8-012-9674-x Phys Chem Chem Phys. https ://doi.org/10.1039/C5CP0 3781F Theillet F-X, Rose HM, Liokatis S, Binolfi A, Thongwichian R, Stuiver Schmitz C, John M, Park AY, Dixon NE, Otting G, Pintacuda G, Huber M, Selenko P (2013) Site-specific NMR mapping and time- T (2006) Efficient chi-tensor determination and NH assignment resolved monitoring of serine and threonine phosphorylation in of paramagnetic proteins. J Biomol NMR 35:79–87. https ://doi. reconstituted kinase reactions and mammalian cell extracts. Nat org/10.1007/s1085 8-006-9002-4 Protoc 8:1416–1432. https ://doi.org/10.1038/nprot .2013.083 Schmitz C, Stanton-Cook MJ, Su X-C, Otting G, Huber T (2008) van der Walt S, Colbert SC, Varoquaux G (2011) The NumPy array: Numbat: an interactive software tool for fitting Deltachi-tensors a structure for efficient numerical computation. Comput Sci Eng to molecular coordinates using pseudocontact shifts. J Biomol 13:22–30. https ://doi.org/10.1109/MCSE.2011.37 NMR 41:179–189. https ://doi.org/10.1007/s1085 8-008-9249-z Vranken WF, Boucher W, Stevens TJ, Fogh RH, Pajon A, Llinas M, Sengupta I, Bhate SH, Das R, Udgaonkar JB (2017) Salt-mediated Ulrich EL, Markley JL, Ionides J, Laue ED (2005) The CCPN oligomerization of the mouse prion protein monitored by real- data model for NMR spectroscopy: development of a software time NMR. J Mol Biol 429:1852–1872. https://doi.or g/10.1016/j. pipeline. Proteins Struct Funct Genet 59:687–696. h t tp s : / /d o i . jmb.2017.05.006 org/10.1002/prot.20449 Skinner SP, Moshev M, Hass MAS, Keizers PHJ, Ubbink M (2013) Wassenaar T, Dijk M, Loureiro-Ferreira N, Schot G, Vries SJ, Schmitz PARAssign—paramagnetic NMR assignments of protein nuclei C, Zwan J, Boelens R, Giachetti A, Ferella L, Rosato A, Bertini on the basis of pseudocontact shifts. J Biomol NMR 55:379–389. I, Herrmann T, Jonker HR, Bagaria A, Jaravine V, Güntert P, https ://doi.org/10.1007/s1085 8-013-9722-1 Schwalbe H, Vranken WF, Doreleijers JF, Vriend G, Vuister GW, Skinner SP, Fogh RH, Boucher W, Ragan TJ, Mureddu LG, Vuister Franke D, Kikhney A, Svergun DI, Fogh RH, Ionides J, Laue ED, GW (2016) CcpNmr AnalysisAssign: a flexible platform for inte- Spronk C, Jurkša S, Verlato M, Badoer S, Pra SD, Mazzucato M, grated NMR analysis. J Biomol NMR 66:111–124. https ://doi. Frizziero E, Bonvin AMJJ. (2012) WeNMR: structural biology org/10.1007/s1085 8-016-0060-y on the grid. J Grid Comput 10:743–767. https ://doi.org/10.1007/ Sormanni P, Piovesan D, Heller GT, Bonomi M, Kukic P, Camilloni s1072 3-012-9246-z C, Fuxreiter M, Dosztanyi Z, Pappu RV, Babu MM, Longhi S, Williamson MP (2013) Using chemical shift perturbation to charac- Tompa P, Dunker AK, Uversky VN, Tosatto SCE, Vendruscolo terise ligand binding. Prog Nucl Magn Reson Spectrosc 73:1–16 M (2017) Simultaneous quantification of protein order and disor - Zimmerman DE, Kulikowski CA, Huang Y, Feng W, Tashiro M, der. Nat Chem Biol 13:339–342. https ://doi.org/10.1038/nchem Shimotakahara S, Chien C, Powers R, Montelione GT (1997) bio.2331 Automated analysis of protein NMR assignments using methods Teilum K, Kunze MBA, Erlendsson S, Kragelund BB (2017) (S)Pin- from artificial intelligence. J Mol Biol 269:592–610. https ://doi. ning down protein interactions by NMR. Protein Sci 26:436–451. org/10.1006/jmbi.1997.1052 https ://doi.org/10.1002/pro.3105 1 3

Journal

Journal of Biomolecular NMRSpringer Journals

Published: May 11, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off