A massive proteogenomic screen identifies thousands of novel human protein coding sequences

Accurate annotation of genes in the human genome is fundamental for biomedical research and genomic data interpretation. The Ensembl, RefSeq, and GENCODE consortiums continuously update the human genome annotations based on new computational and experimental evidence, and new proteins were identified constantly. The Genotype-Tissue Expression (GTEx) project has generated more than 15,000 RNA sequencing dataset from multiple-tissues of more than 800 donors which allows to model almost all transcripts and proteins in the human genome. Using proteins translated from the GTEx transcript model, more than 21 million in-silico trypsin-digested peptides were generated. To identify high-confidence novel proteins with proteomic support, we screened more than 2,000 proteomic projects in the PRIDE database and selected more than 50,000 mass spectrometry (MS) runs from 923 projects. These MS data were used to validate the predicted novel peptides. With a stringent standard, we identified almost 20,000 novel peptides. This dataset include files used in the the above analysis. More details can be found in the GitHub page (https://github.com/ATPs/human_novo_protein_2022)