The shrinking human protein coding complement: are there fewer than 20,000 genes?

Determining the full complement of protein-coding genes is a key goal of genome annotation. The most powerful approach for confirming protein coding potential is the detection of cellular protein expression through peptide mass spectrometry experiments. Here we map the peptides detected in 7 large-scale proteomics studies to almost 60% of the protein coding genes in the GENCODE annotation the human genome. We find that conservation across vertebrate species and the age of the gene family are key indicators of whether a peptide will be detected in proteomics experiments. We find peptides for most highly conserved genes and for practically all genes that evolved before bilateria. At the same time there is almost no evidence of protein expression for genes that have appeared since primates, or for genes that do not have any protein-like features or cross-species conservation. We identify 19 non-protein-like features such as weak conservation, no protein features or ambiguous annotations in major databases that are indicators of low peptide detection rates. We use these features to describe a set of 2,001 genes that are potentially non-coding, and show that many of these genes behave more like non-coding genes than protein-coding genes. We detect peptides for just 3% of these genes. We suggest that many of these 2,001 genes do not code for proteins under normal circumstances and that they should not be included in the human protein coding gene catalogue. These potential non-coding genes will be revised as part of the ongoing human genome annotation effort.