MaxQuant goes Linux

MaxQuant goes Linux correspondence 100% To the Editor: We report a Linux version (https://chorusproject.org). We expect that of MaxQuant (http://www.biochem. the number of applications will increase 1,000 mpg.de/5111795/maxquant), our popular with our Linux-compatible MaxQuant software platform for the analysis of shotgun version. We envision that proteomics 61% proteomics data. core facilities, for instance, will benefit Write output tables (52%) LFQ normalization (94%) One of our main intentions in developing from the combination of command-line Protein assembly (118%) MaxQuant was to ‘take the pain out of ’ access and Linux compatibility, which Second peptide search (23%) quantifying large collections of protein enables standardized high-throughput Main search (31%) profiles . However, unlike, for instance, data analysis. The MaxQuant code base is Mass recalibration (41%) First search (61%) the Trans-Proteomic Pipeline , the original identical for Windows and for Linux; thus Feature detection (44%) version of MaxQuant could be run only on .NET on Windows Mono on Linux there is only a single distributable running Microsoft Windows, and thus its use was on both operating systems, which can be Fig. 1 | Benchmarking MaxQuant on Linux and restricted in high-performance computing downloaded from http://www.maxquant. Windows. We analyzed 300 LC-MS runs with environments, which very rarely use org (version 1.6.1.0). MaxQuant is freeware, MaxQuant using 120 logical cores in parallel, once Windows as an operating system. When and contributions to new functionality with Ubuntu Linux (version 16.04.3) and once we began developing MaxQuant, Windows are collaboration-based. The code of open with Windows server 2012 R2 as the operating was the only operating system supported source parts is available at https://github. system. We used identical hardware in both by vendor-provided raw data access com/JurgenCox/compbio-base. ❐ cases: four Intel Xeon E7-4870 CPUs and 256 libraries. Therefore, we wrote MaxQuant GB of DDR3 RAM. The total running times are in the C# programming language on top Pavel Sinitcyn, Shivani Tiwary, Jan Rudolph, shown, and several long-running sub-workflows of the Windows-only .NET framework. Petra Gutenbrunner, Christoph Wichmann, are highlighted. Percentages indicate the amount Windows support for cloud platforms is Şule Yılmaz, Hamid Hamzeiy, Favio Salinas of time needed to complete the relevant process more expensive, and the operating system and Jürgen Cox* in Linux as a percentage of the total time required is harder to use and less scalable compared Computational Systems Biochemistry, Max Planck for the same process in Windows. with Linux. Institute for Biochemistry, Martinsried, Germany. We recently carried out a major *e-mail: cox@biochem.mpg.de restructuring of the MaxQuant codebase, Published online: 31 May 2018 and we made it compatible with Mono MaxQuant run on 120 logical cores, we https://doi.org/10.1038/s41592-018-0018-y (https://www.mono-project.com/), an observed that the Linux version showed alternative cross-platform implementation highly superior parallelization performance, of the .NET framework. Furthermore, we with speed 64% faster than that observed References 1. Cox, J. & Mann, M. Nat. Biotechnol. 26, 1367–1372 (2008). now provide an entry point to MaxQuant under a Windows server operating system 2. Azvolinsky, A., DeFrancesco, L., Waltz, E. & Webb, S. Nat. from the command line without the using identical hardware. MaxQuant uses Biotechnol. 34, 256–261 (2016). need to start its graphical user interface, operating system processes, rather than the 3. Deutsch, E. W. et al. Proteomics Clin. Appl. 9, 745–754 (2015). 4. Judson, B., McGrath, G., Peuchen, E. H., Champion, M. M. & which allows execution from scripts intrinsic multi-threading mechanism of C#, Brenner, P. In Proc. 8th Workshop on Scientic Cl fi oud Computing or other processing tools. Meanwhile, to realize parallel execution, and it manages (eds. Chard, K. et al.) 17–24 (ACM, New York, 2017). Thermo Fisher Scientific has released the load-balancing of an arbitrarily large set its platform-independent and Mono- of raw data files over a specified number of compatible implementation of its raw data processors by itself. We hypothesize that this Acknowledgements This project has received funding from the European access library (http://planetorbitrap.com/ allows Linux to optimize parallel execution Union’s Horizon 2020 research and innovation program rawfilereader), and hopefully more vendors to the high extent that we observed. A larger (grant agreement no. 686547 to J.C., J.R. and S.Y.) and will follow soon. Together, this leads to a benchmark study is under way, in which from the FP7 (grant GA ERC-2012-SyG_318987–ToPAG situation in which large-scale computing of we will investigate the dependence of the to S.T. and F.S.). proteomics data with MaxQuant becomes increased speed on hardware such feasible on all common platforms. as, for instance, the type of CPU and Author contributions P.S., S.T., J.R., P.G., C.W., S.Y., H.H., F.S. and J.C. developed When we parallelized the MaxQuant storage systems. the software. P.S. conducted the performance analysis. workflow over only a few central processing MaxQuant has already been adapted J.C. wrote the manuscript. unit (CPU) cores, we hardly noticed in several forms for cloud and high- a difference in performance between performance computing applications, Linux and Windows (Fig. 1). However, as described, for instance, by Competing interests in benchmarking of a highly parallelized Judson et al. and on the Chorus platform The authors declare no competing interests. Nature Methods | VOL 15 | JUNE 2018 | 401 | www.nature.com/naturemethods © 2018 Nature America Inc., part of Springer Nature. All rights reserved. Running time (min) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Nature Methods Springer Journals

Loading next page...
1 Page
 
/lp/springer_journal/maxquant-goes-linux-WnosdIDORC
Publisher
Springer Journals
Copyright
Copyright © 2018 by The Author(s)
Subject
Life Sciences; Life Sciences, general; Biological Techniques; Biological Microscopy; Biomedical Engineering/Biotechnology; Bioinformatics; Proteomics
ISSN
1548-7091
eISSN
1548-7105
D.O.I.
10.1038/s41592-018-0018-y
Publisher site
See Article on Publisher Site

Abstract

correspondence 100% To the Editor: We report a Linux version (https://chorusproject.org). We expect that of MaxQuant (http://www.biochem. the number of applications will increase 1,000 mpg.de/5111795/maxquant), our popular with our Linux-compatible MaxQuant software platform for the analysis of shotgun version. We envision that proteomics 61% proteomics data. core facilities, for instance, will benefit Write output tables (52%) LFQ normalization (94%) One of our main intentions in developing from the combination of command-line Protein assembly (118%) MaxQuant was to ‘take the pain out of ’ access and Linux compatibility, which Second peptide search (23%) quantifying large collections of protein enables standardized high-throughput Main search (31%) profiles . However, unlike, for instance, data analysis. The MaxQuant code base is Mass recalibration (41%) First search (61%) the Trans-Proteomic Pipeline , the original identical for Windows and for Linux; thus Feature detection (44%) version of MaxQuant could be run only on .NET on Windows Mono on Linux there is only a single distributable running Microsoft Windows, and thus its use was on both operating systems, which can be Fig. 1 | Benchmarking MaxQuant on Linux and restricted in high-performance computing downloaded from http://www.maxquant. Windows. We analyzed 300 LC-MS runs with environments, which very rarely use org (version 1.6.1.0). MaxQuant is freeware, MaxQuant using 120 logical cores in parallel, once Windows as an operating system. When and contributions to new functionality with Ubuntu Linux (version 16.04.3) and once we began developing MaxQuant, Windows are collaboration-based. The code of open with Windows server 2012 R2 as the operating was the only operating system supported source parts is available at https://github. system. We used identical hardware in both by vendor-provided raw data access com/JurgenCox/compbio-base. ❐ cases: four Intel Xeon E7-4870 CPUs and 256 libraries. Therefore, we wrote MaxQuant GB of DDR3 RAM. The total running times are in the C# programming language on top Pavel Sinitcyn, Shivani Tiwary, Jan Rudolph, shown, and several long-running sub-workflows of the Windows-only .NET framework. Petra Gutenbrunner, Christoph Wichmann, are highlighted. Percentages indicate the amount Windows support for cloud platforms is Şule Yılmaz, Hamid Hamzeiy, Favio Salinas of time needed to complete the relevant process more expensive, and the operating system and Jürgen Cox* in Linux as a percentage of the total time required is harder to use and less scalable compared Computational Systems Biochemistry, Max Planck for the same process in Windows. with Linux. Institute for Biochemistry, Martinsried, Germany. We recently carried out a major *e-mail: cox@biochem.mpg.de restructuring of the MaxQuant codebase, Published online: 31 May 2018 and we made it compatible with Mono MaxQuant run on 120 logical cores, we https://doi.org/10.1038/s41592-018-0018-y (https://www.mono-project.com/), an observed that the Linux version showed alternative cross-platform implementation highly superior parallelization performance, of the .NET framework. Furthermore, we with speed 64% faster than that observed References 1. Cox, J. & Mann, M. Nat. Biotechnol. 26, 1367–1372 (2008). now provide an entry point to MaxQuant under a Windows server operating system 2. Azvolinsky, A., DeFrancesco, L., Waltz, E. & Webb, S. Nat. from the command line without the using identical hardware. MaxQuant uses Biotechnol. 34, 256–261 (2016). need to start its graphical user interface, operating system processes, rather than the 3. Deutsch, E. W. et al. Proteomics Clin. Appl. 9, 745–754 (2015). 4. Judson, B., McGrath, G., Peuchen, E. H., Champion, M. M. & which allows execution from scripts intrinsic multi-threading mechanism of C#, Brenner, P. In Proc. 8th Workshop on Scientic Cl fi oud Computing or other processing tools. Meanwhile, to realize parallel execution, and it manages (eds. Chard, K. et al.) 17–24 (ACM, New York, 2017). Thermo Fisher Scientific has released the load-balancing of an arbitrarily large set its platform-independent and Mono- of raw data files over a specified number of compatible implementation of its raw data processors by itself. We hypothesize that this Acknowledgements This project has received funding from the European access library (http://planetorbitrap.com/ allows Linux to optimize parallel execution Union’s Horizon 2020 research and innovation program rawfilereader), and hopefully more vendors to the high extent that we observed. A larger (grant agreement no. 686547 to J.C., J.R. and S.Y.) and will follow soon. Together, this leads to a benchmark study is under way, in which from the FP7 (grant GA ERC-2012-SyG_318987–ToPAG situation in which large-scale computing of we will investigate the dependence of the to S.T. and F.S.). proteomics data with MaxQuant becomes increased speed on hardware such feasible on all common platforms. as, for instance, the type of CPU and Author contributions P.S., S.T., J.R., P.G., C.W., S.Y., H.H., F.S. and J.C. developed When we parallelized the MaxQuant storage systems. the software. P.S. conducted the performance analysis. workflow over only a few central processing MaxQuant has already been adapted J.C. wrote the manuscript. unit (CPU) cores, we hardly noticed in several forms for cloud and high- a difference in performance between performance computing applications, Linux and Windows (Fig. 1). However, as described, for instance, by Competing interests in benchmarking of a highly parallelized Judson et al. and on the Chorus platform The authors declare no competing interests. Nature Methods | VOL 15 | JUNE 2018 | 401 | www.nature.com/naturemethods © 2018 Nature America Inc., part of Springer Nature. All rights reserved. Running time (min)

Journal

Nature MethodsSpringer Journals

Published: May 31, 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