Loose ends: almost one in five human genes still have unresolved coding status.

Area of Science:

• Genomics
• Proteomics
• Bioinformatics

Background:

• The human genome was sequenced 17 years ago, yet the human proteome remains under revision.
• Discrepancies exist across major databases (Ensembl/GENCODE, RefSeq, UniProtKB) regarding gene annotations.
• One in eight genes are annotated differently among these key reference sets.

Purpose of the Study:

• To investigate genes with conflicting annotations across major human genome databases.
• To identify potentially misclassified protein-coding genes within the human reference proteome.
• To assess the impact of annotation discrepancies on large-scale biomedical research.

Main Methods:

• In-depth investigation of 2764 genes with conflicting coding/non-coding annotations.
• Analysis of large-scale genetic variation data to assess purifying selection.
• Evaluation of transcript and protein evidence for gene candidates.
• Comparison of evolutionary patterns (neutral evolution) for gene sets.

Main Results:

• Most of the 2764 investigated genes show no protein-like purifying selection, suggesting they do not code for functional proteins.
• An additional 1470 genes, annotated as coding in all three databases, exhibit characteristics of non-coding genes or pseudogenes.
• These 1470 genes display neutral evolution and possess less supporting transcript/protein evidence compared to validated coding genes.
• The study estimates an overestimation of at least 2000 coding genes in current human reference databases.

Conclusions:

• Current human reference databases likely overestimate the number of coding genes by over 2000.
• Misclassified non-coding genes introduce noise and complexity into large-scale biomedical experiments.
• Accurate determination of the human reference proteome is crucial for fundamental biological research and biomedical projects.