Identification and Validation of Human Missing Proteins and Peptides in Public Proteome Databases: Data Mining Strategy

Abstract

In an attempt to complete human proteome project (HPP), Chromosome-Centric Human Proteome Project (C-HPP) launched the journey of missing protein (MP) investigation in 2012. However, 2579 and 572 protein entries in the neXtProt (2017-1) are still considered as missing and uncertain proteins, respectively. Thus, in this study, we proposed a pipeline to analyze, identify, and validate human missing and uncertain proteins in open-access transcriptomics and proteomics databases. Analysis of RNA expression pattern for missing proteins in Human protein Atlas showed that 28% of them, such as Olfactory receptor 1I1 ( O60431 ), had no RNA expression, suggesting the necessity to consider uncommon tissues for transcriptomic and proteomic studies. Interestingly, 21% had elevated expression level in a particular tissue (tissue-enriched proteins), indicating the importance of targeting such proteins in their elevated tissues. Additionally, the analysis of RNA expression level for missing proteins showed that 95% had no or low expression level (0-10 transcripts per million), indicating that low abundance is one of the major obstacles facing the detection of missing proteins. Moreover, missing proteins are predicted to generate fewer predicted unique tryptic peptides than the identified proteins. Searching for these predicted unique tryptic peptides that correspond to missing and uncertain proteins in the experimental peptide list of open-access MS-based databases (PA, GPM) resulted in the detection of 402 missing and 19 uncertain proteins with at least two unique peptides (≥9 aa) at <(5 × 10)% FDR. Finally, matching the native spectra for the experimentally detected peptides with their SRMAtlas synthetic counterparts at three transition sources (QQQ, QTOF, QTRAP) gave us an opportunity to validate 41 missing proteins by ≥2 proteotypic peptides.

Progress on Identifying and Characterizing the Human Proteome: 2018 Metrics from the HUPO Human Proteome Project

“…The editorial for the JPR 2017 special issue 12 highlighted six papers that used a variety of promising methods to find Missing Proteins, with a total of 32 identified for validation as PE1: 15 from Li et al 13 using Triton X-100 solubilization plus ProteoMiner hexapeptide-covered beads as an enrichment/equalization strategy for low-abundance proteins with kidney, bladder, liver, and colorectal specimens, and confirmation with PRM; 12 from Carapito et al 14 on the sperm proteome with PRM confirmation; 3 from Wang et al 15 using a multi-protease strategy on testis; 1 from Peng 16 from the kidney phosphoproteome; 1 from Meyfour et al 17 based on biological studies of a Y chromosome protein in cardiac development; plus 41 cautiously forwarded from Elguoshy et al 18 using the stranded peptides approach described below. Twenty MP reported by Li et al 8 , including 5 without PRM validation that had not been counted in the editorial, are now classified as canonical in PeptideAtlas and validated as PE1 in neXtProt.…”

“…PeptideAtlas, which several years ago developed PASSEL for extensive SRM data, has now begun to process PRM data. The second development was the strategy of the Chromosome 7 team (Baker et al 20 ) and the Chromosome X team (Elguoshy et al 18 ) to identify what we have named “stranded peptides” in major databases that could be combined to make a pair of proteotypic peptides for individual missing proteins and then use the reported spectra from the original work (preferably deposited in PRIDE or other accessible repository with a PXD identifier) to compare with the spectra available in SRMAtlas 21 for the corresponding synthetic peptide. This work will be facilitated by development of a Universal Spectrum Identifier, soon to be released by PeptideAtlas and PSI.…”

Biochemical Tools for Tracking Proteolysis

“…In particular, three and seven members of the ADAM and ADAMTS (a disintegrin and metalloproteinase with thrombospondin motifs) are reported not having evidence at the protein level. Recent work by Elguoshy, Yamamoto, and co-workers , aim at validating these missing proteins by using bioinformatics and SRMAtlas synthetic peptides to find these proteins in proteomics data sets. We cross-referenced the 72 missing proteases and inhibitors with their published data, but while 21 were initially considered good candidates, none passed the final validation threshold.…”

A high-stringency blueprint of the human proteome

“…This assay library enables targeted identification and quantification of a theoretical maximum of 99.7% of the human proteome 36 , provided proteins are expressed spatiotemporally at concentrations amenable to MS detection. For example, SRMAtlas supported the C-HPP Chr X team’s confident identification of missing proteins 37 .…”