Dataset Information

De novo nine-species benchmark

ABSTRACT: We created a new version of the nine-species benchmark originally described by Tran et al. [Tran2017]. To do so, we downloaded the RAW files from the same nine PRIDE projects (PXD005025, PXD004948, PXD004325, PXD004565, PXD004536, PXD004947, PXD003868, PXD004467, PXD004424) and converted them to MGF format using ThermoRawFileParser v1.3.4. We also downloaded the corresponding nine Uniprot reference proteomes and constructed a Tide index for each one, using Crux version 4.1. For one species (Vigna mungo), no reference proteome is available, so we used the proteome of the closely related species Vigna radiata. We allowed for the following variable modifications: Met oxidation, Asn deamidation, Gln deamidation, N-term acetylation, N-term carbamylation, N-term NH3 loss, and the combination of N-term carbamylation and NH3 loss by using the tide-index options "--mods-spec 1M+15.994915, 1N+0.984016, 1Q+0.984016 --nterm-peptide-mods-spec 1X+42.010565, 1X+43.005814, 1X-17.026549, 1X+25.980265 --max-mods 3". Note that one of the nine experiments (Mus musculus) was performed using SILAC labeling, but we searched without SILAC modifications and hence include in the benchmark only PSMs from unlabeled peptides. Each index also contains a shuffled decoy peptide corresponding to each target peptide. Each MGF file was searched against the corresponding index using the precursor window size and fragment bin tolerance specified in the original study. We used XCorr scoring with Tailor calibration, and we allowed for 1 isotope error in the selection of candidate peptides. All search results were then analyzed jointly per species using the Crux implementation of Percolator, with default parameters. For the benchmark, we retained all PSMs with Percolator q value < 0.01. We identified 13 MGF files with fewer than 100 accepted PSMs, and we eliminated all of these PSMs from the benchmark. We then post-processed the PSMs to eliminate peptides that are shared between species. Among the 229,984 unique peptides, we identified 3797 (1.7%) that occur in more than one species. For each such peptide, we selected one of the associated species at random and then eliminated all PSMs containing that peptide in other species. The final benchmark dataset consists of 2.8 million PSMs drawn from 343 RAW files, exported as annotated MGF files. Note: the initial data submission contained annotated MGF files without considering the N-terminal modifications mentioned above. The update available in the `/MSV000090982/updates/2024-05-14_woutb_71950b89/peak/9speciesbenchmark` FTP directory contains the corrected MGF files that are directly compatible with Casanovo. [Tran2017] Tran, N. H., Zhang, X., Xin, L., Shan, B. & Li, M. De novo peptide sequencing by deep learning. Proceedings of the National Academy of Sciences of the United States of America 31, 8247-8252 (2017).

INSTRUMENT(S): Q Exactive

ORGANISM(S): Bacillus Subtilis (ncbitaxon:1423) Candidatus Thiodiazotropha Endoloripes (ncbitaxon:1818881) Vigna Mungo (ncbitaxon:3915) Solanum Lycopersicum (ncbitaxon:4081) Saccharomyces Cerevisiae (ncbitaxon:4932) Homo Sapiens (ncbitaxon:9606) Mus Musculus (ncbitaxon:10090) Methanosarcina Mazei (ncbitaxon:2209) Apis Mellifera (ncbitaxon:7460)

SUBMITTER: William Stafford Noble

PROVIDER: MSV000090982 | MassIVE | Mon Jan 02 03:05:00 GMT 2023

REPOSITORIES: MassIVE

ACCESS DATA

Dataset's files

Source:

			Action	DRS
		Other

Items per page:

1 - 1 of 1

Similar Datasets

Project description:In this project, we aim to pair-wise analyze the genomes, transcriptomes and proteomes of in-bred rats originating from two different genetic backgrounds. These two strains are Brown Norway (BN-Lx) and Spontaneously Hypertensive Rats (SHR). First, we re-sequenced the genomes for both BN and SHR rats, followed by RNA-seq and proteomics of their liver tissues. We then append novel predicted gene models, non-synonymous SNPs and INDELs (derived from genome re-sequencing), as well as transcript variants such as RNA-editing and alternative splicing (derived from RNA-seq) that can diversify existing protein sequences onto the ENSEMBL rat FASTA (Build 68) to build an enhanced database. For proteomics studies, equal amount of liver lysates were digested with trypsin, LysC, GluC, AspN and chymotrypsin and were individually fractionated with strong cationic exchange chromatography. Doubly- and triply-charged fractions were analyzed with an Triple-TOF 5600 with collision-activated dissociation (CAD); while electron-transfer dissociation (ETD) was applied for fractions containing triple charges and above with a LTQ-Orbitrap Velos. Data analysis: Peak List generation: For Wiff files generated from TripleTOF 5600, tandem MS spectra were de-isotoped, charge- deconvoluted and peak lists converted to Mascot generic format (MGF) files using AB Sciex Data Converter (version 1.1). For data generated from the LTQ-Orbitrap Velos, Raw files were converted to MGF files using Proteome Discoverer (version 1.3). The non-fragment filter was used to simplify ETD spectra and the Top N filter for the HCD spectra. Three MGF files were generated (one for HCD, one for ETD IT and one for ETD FT). The files with an orbitrap readout were deisotoped and charge de-convoluted. Database Searching: All MGF files were queried with Mascot search engine (version 2.3) via Proteome Discoverer version 1.3 (PD 1.3, Thermo Fisher) for submission. The spectra were searched against in-house database (NGS_COMBINED). One of the five different enzymes used (Trypsin/P, LysC/P, Chymotrypsin, GluC-DE and AspN_ambic) were selected for each file and up to 9 missed cleavages were allowed. Cysteine carbamidomethylation was set as fixed modification, and oxidation of methionine and acetylation of the N-term as variable modifications. Peptide tolerance was initially set to 50 ppm and the MS/MS tolerance was set to 0.1 Da (for TOF readout), 0.02 Da (orbitrap readout) and 0.5 Da (ion trap readout). All peptide-spectrum matches (PSMs) were evaluated with Percolator for validation. We classified each PSM based on their q value. For proteins identification, we used set a high stringency filter of q = 0 (0% FDR). For peaks lists that do not yield any peptide matches, we exported them with PD 1.3 for further analysis. De novo search with PEAKS: Unassigned peak lists that are exported were re-analyzed with another software suite i.e. PEAKS Studio (version 6.0). The identification workflows is as follows. Peak lists were first filtered with a quality value of 0.65 as suggested by the manufacturer followed by de novo spectra interpretation. In this step, both peptide tolerance and MS/MS tolerance were set according to MASCOT search. To broaden the search space for these unassigned spectra, we additionally set de-amidation of asparagine and glutamine, and pyro-glu from glutamic acid and glutamine as variable modifications, on top of the other modifications indicated above. Maximum allowed variable PTM per peptide was set to 3. Finally de novo interpreted PSMs were submitted to PEAKS DB database matching, this time allowing semi-enzymatic specificity and a maximum cleavages per peptide of 2. Database used was set to NGS_COMBINED. FDR was estimated using decoy-fusion. The genomics and transcriptomics data are already deposited in the respective EBI repositories. Some of these data are derived from an already published manuscript. For the genomics data (from: Genetic basis of transcriptome differences between the founder strains of the rat HXB/BXH recombinant inbred panel by Simonis et al PMID:22541052) DNA data in Sequence Read Archive (SRA): BN-Lx genome: ERP001355 http://www.ebi.ac.uk/ena/data/view/ERP001355, SHR genome: ERP001371, BN reference genome: ERP000510, http://www.ebi.ac.uk/ena/data/view/ERP000510. RNA data in ArrayExpress: BN-Lx and SHR fragment RNA-seq data: E-MTAB-1029 http://www.ebi.ac.uk/arrayexpress/experiments/E-MTAB-1029, BN-Lx and SHR paired-end RNA-seq data: to be submitted.

Dataset Information

De novo nine-species benchmark

Dataset's files

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets