All files for ProteomeXchange ID PXD001010 were downloaded, along with the forward and reversed FASTA and mzXML files (n = 46) used in the library peptide search (requested of the authors). Using Skyline (daily version 3.7.1.11571) first a spectral library was built using the iProphet score cut-off 0.0242 suggested in the paper as achieving 1% FDR at the PSM level, with 1031 ambiguously matched peptides excluded, resulting in 82,439 unique peptides (104,993 entries). Because the Biognosys iRT standard peptides used were not included in the search, an iRT library was created from these files by adding all detected peptides with the iRT standards added as targets in Skyline and importing the mzXML files for MS1 extraction. The iRT values were then calculated using both the extracted peaks for the iRT standards and target peptide MS1 peaks where the peak contained a matching MS/MS ID, because the runs used fractionated samples and all peptides were not expected in all runs.
For DIA, allowing unique peptides of length 7 to 45 resulting from semi-tryptic cleavage with up to 2 missed cleavages, with Carbamidomethyl (C) and optionally Oxidation (M), precursors of charge states 2, 3, or 4, from 400-1200 m/z (the range covered by the DIA method), with 6 product transitions found in the library, of y or b type and 1 or 2 charge state (excluding y1, y2, b1, and b2). Chromatogram extraction was set to use TOF extraction at 18,000 resolving power with high-selectivity extraction applying to all MS/MS spectra within 10 minutes of predicted retention times using the iRT library.
Importing the FASTA file and then removing duplicate peptides and empty proteins resulted in targets for 4,603 proteins, 68,910 peptides, 87,042 precursors, and 522,252 fragment ions, at 32% protein, 2.7% unique peptide FDR by reversed sequence decoy counting (decoys/targets). The protein FDR is likely overstated because the FASTA file contains only 6717 protein sequences, which means as many as 2/3 of false peptides can be expected to occur in a true protein, while the same is not true for detections of reversed peptides. Even at 10% protein FDR, however, this target set seemed to contain a higher error rate than we felt desirable.
For these experiments, we decided to rebuild the library using the iProphet score cut-off 0.9, with 361 ambiguously matched peptides excluded, resulting in 64,501 unique peptides (84,245 entries). For our most inclusive method, we chose to include only fully-tryptic peptides and no variable modifications (dropping Oxidation (M)), which resulted in targets for 4152 proteins, 36,889 peptides, 48,082 precursors, and 288,492 fragment ions, at 2.6% protein, 0.29% unique peptide FDR by reversed sequence decoy. counting. An equal number of shuffled sequence decoys were generated for mProphet model generation.
The 18 runs were then imported into the template an mProphet model trained and applied. The MSstats report was exported for further analysis. The differential abundance analysis was performed by MSstats (v3.16.0) R package. Details for data processing and statistical analysis are available in description.pdf ('Methods' folder).
**Publication : Choi et al. (Under revision) MSstats increases the reproducibility of detecting differentially abundant proteins across tools that process raw mass spectra.
[doi:10.25345/C5114P]
[See
results attachment job
for details]
Keywords: MassIVE.quant reviewed - Platinum
Number of Files: | |
Total Size: | |
Experimental Design | |
Conditions:
|
|
Biological Replicates:
|
|
Technical Replicates:
|
|
Identification Results | |
Proteins (Human, Remapped):
|
|
Proteins (Reported):
|
|
Peptides:
|
|
Variant Peptides:
|
|
PSMs:
|
|
Quantification Results | |
Differential Proteins:
|
|
Quantified Proteins:
|
|
Browse Reanalysis Files | |
Browse Quantification Results | Browse Metadata |
FTP Download Link (click to copy):
|