Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2021 Dec;39(12):1563-1573.
doi: 10.1038/s41587-021-00968-7. Epub 2021 Jul 8.

MaxDIA enables library-based and library-free data-independent acquisition proteomics

Pavel Sinitcyn #  1 , Hamid Hamzeiy #  1 , Favio Salinas Soto #  1 , Daniel Itzhak  2 , Frank McCarthy  2 , Christoph Wichmann  1 , Martin Steger  3 , Uli Ohmayer  3 , Ute Distler  4 , Stephanie Kaspar-Schoenefeld  5 , Nikita Prianichnikov  1 , Şule Yılmaz  1 , Jan Daniel Rudolph  1   6 , Stefan Tenzer  4 , Yasset Perez-Riverol  7 , Nagarjuna Nagaraj  5 , Sean J Humphrey  8 , Jürgen Cox  9   10
Affiliations

MaxDIA enables library-based and library-free data-independent acquisition proteomics

Pavel Sinitcyn et al. Nat Biotechnol. 2021 Dec.

Abstract

MaxDIA is a software platform for analyzing data-independent acquisition (DIA) proteomics data within the MaxQuant software environment. Using spectral libraries, MaxDIA achieves deep proteome coverage with substantially better coefficients of variation in protein quantification than other software. MaxDIA is equipped with accurate false discovery rate (FDR) estimates on both library-to-DIA match and protein levels, including when using whole-proteome predicted spectral libraries. This is the foundation of discovery DIA-hypothesis-free analysis of DIA samples without library and with reliable FDR control. MaxDIA performs three- or four-dimensional feature detection of fragment data, and scoring of matches is augmented by machine learning on the features of an identification. MaxDIA's bootstrap DIA workflow performs multiple rounds of matching with increasing quality of recalibration and stringency of matching to the library. Combining MaxDIA with two new technologies-BoxCar acquisition and trapped ion mobility spectrometry-both lead to deep and accurate proteome quantification.

PubMed Disclaimer

Conflict of interest statement

The authors state that they have potential conflicts of interest regarding this work: M.S. and U.O. are employees of Evotec, N.N. and S.K.S. are employees of Bruker and J.D.R. is an employee of Bosch. The remaining authors declare no competing financial interests.

Figures

Fig. 1
Fig. 1. Overview of the MaxDIA workflow.
MaxDIA can be operated in library and discovery mode. Many concepts and algorithms—for instance, for protein quantification—are re-used from the conventional MaxQuant workflow for DDA data and have been further developed for DIA. This results in an end-to-end DIA software that contains many established MaxQuant concepts, such as label-free quantification with MaxLFQ or iBAQ quantification. RT, retention time.
Fig. 2
Fig. 2. 3D/4D feature detection of precursors and fragments.
a, Visualization of precursors and fragments of a peptide measured on an Orbitrap. The raw data can be visualized together with the peak detection results as heat maps and 3D models for precursor and fragment data in the graphical user interface of MaxQuant. b, Two peptides with nearly equal mass, both with charge 2 and having very similar retention times, are resolved by ion mobility on a timsTOF Pro mass spectrometer. A heat map visualizes intensities as a function of retention time and collision cross-section for the precursor isotope patterns. The two respective MS/MS spectra of fragments assigned to the precursors are shown. RT, retention time.
Fig. 3
Fig. 3. Performance evaluation.
Twenty-seven technical replicates of HepG2 cell lysate were analyzed on an Orbitrap mass spectrometer (Methods). a, Number of identified protein groups with 1% FDR on protein and peptide level and number of peptides at 1% library-to-DIA-sample FDR obtained with MaxDIA, Spectronaut 13 and Spectronaut 14. b, Histograms of peptide lengths identified with MaxDIA (blue) and Spectronaut 13 (red). c, Number of proteins with, at most, x out of 27 valid values for Spectronaut 13 (red), Spectronaut 14 (magenta) and MaxDIA with MaxLFQ minimum ratio count = 1 (blue, dashed) and = 2 (blue, solid). Multiple curves for the two MaxQuant series of curves correspond to seven different choices for the transfer q value (0.01, 0.03, 0.05, 0.1, 0.3, 0.5 and 1). d, Histograms of coefficients of variation for analyses with default settings in MaxDIA (solid blue) and in Spectronaut 13 and Spectronaut 14 (open). e, log–log scatter plot of LFQ intensities between two representative replicates obtained with MaxQuant. The two replicates were chosen to have the median Pearson correlation of all pairwise replicate comparisons. f, Same as in e for Spectronaut intensities. Similarly, the two replicates were chosen to represent the median Pearson correlation coefficient of all pairwise comparisons. g, Heat map with all pairwise Pearson correlations among the 27 replicates for MaxDIA (upper triangle) and Spectronaut (lower traingle). The two values corresponding to the comparisons in e and f are marked with red squares. h, log–log scatterplot of iBAQ protein intensities from MaxDIA against Spectronaut protein intsnsities. i, log–log scatterplot of MaxDIA iBAQ values averaged over the replicates against RPKM values from RNA-seq data. j, Same as i with protein intensities from Spectronaut.
Fig. 4
Fig. 4. Internal and external FDR.
a, Number of identifications (blue: matches; green: peptides; red: protein groups) as a function of estimated FDR. The FDR is estimated once with the ‘internal’ target-decoy method implemented in MaxQuant (solid lines) and once with the ‘external’ method using mixing maize and human samples for generating the library and using only human sample in the DIA runs (dashed lines). b, Same as in a but using in silico predicted libraries generated using DeepMass:Prism c, Same as a but using the raw score instead of the machine learning–derived score. d, Same as b but using the raw score instead of the machine learning–derived score.
Fig. 5
Fig. 5. MaxLFQ for DIA.
a, Stacked interquartile rages of protein ratio distributions in the small-ratio four-species dataset from Bruderer et al. using different versions of MaxLFQ for DIA and compared to the results from this publication. MaxDIA is capable of MS1 and MS2 level as well as hybrid quantification modes. b, Quantification of a three-species benchmark mixture measured on a SCIEX TripleTOF 6600 instrument mixing proteomes from three species in defined ratio with MaxLFQ for DIA. The accompanying DDA library was used. The box plots here and in the subsequent panels are based on the numbers of data points given in the tables below the respective plot (valid LFQ ratios). All box plots indicate the median and the first and third quartiles as box ends. Whiskers are positioned 1.5 box lengths away from the box ends. c, Same as b but analyzed with MaxDIA in discovery mode. d, Quantification of a three-species benchmark mixture measured on a Bruker timsTOF Pro instrument mixing proteomes from three species in defined ratio using a DDA library. e, Same as d but analyzed in discovery mode.
Fig. 6
Fig. 6. BoxCar and fractionated DIA.
a, Schedule of libraries and DIA samples. Three different library approaches—single-shot, deep-fractionated and discovery mode—were compared to single-shot, deep-fractionated DIA samples. b, MaxLFQ quantification among three replicates of fractionated BoxCar DIA samples analyzed in discovery DIA mode. All pairwise Pearson correlations are above 0.99. c, Venn diagram-like comparison represented as bar plot between RNA-seq data of HEK cells and three different library methods applied to the fractionated DIA samples. All data have been mapped to gene identifiers d, Histogram of protein identifications mapped to gene identifiers sorted into bins according to log2 RPKM values of the RNA-seq data.
See this image and copyright information in PMC

Comment in

  • Proteomic analysis with MaxDIA.
    Singh A. Singh A. Nat Methods. 2021 Sep;18(9):988. doi: 10.1038/s41592-021-01267-4. Nat Methods. 2021. PMID: 34480153 No abstract available.

Similar articles

See all similar articles

Cited by

See all "Cited by" articles

References

    1. Doerr A. DIA mass spectrometry. Nat. Methods. 2014;12:35–35.
    1. Navarro P, et al. A multicenter study benchmarks software tools for label-free proteome quantification. Nat. Biotechnol. 2016;34:1130–1136. - PMC - PubMed
    1. Cox J, Mann M. MaxQuant enables high peptide identification rates, individualized p.p.b.-range mass accuracies and proteome-wide protein quantification. Nat. Biotechnol. 2008;26:1367–1372. - PubMed
    1. Azvolinsky A, DeFrancesco L, Waltz E, Webb S. 20 years of Nature Biotechnology research tools. Nat. Biotechnol. 2016;34:256–261. - PubMed
    1. Sinitcyn P, Rudolph JD, Cox J. Computational methods for understanding mass spectrometry-based shotgun proteomics. Annu. Rev. Biomed. Data Sci. 2018;1:207–234.

Publication types

-