2026 ECNP Congress

PS01-1204

When the background matters: reference lists selection impacts brain genome-wide association studies and exome enrichment analyses

Komal Zade — B. Timoney, P. Guasoni, S. Bach, D. Tropea · Trinity College Dublin, Department of Psychiatry, Dublin, Ireland

2026 ECNP Congress

Background

High-throughput genetic and molecular profiling routinely produces lists of genes associated with a phenotype, including hits from genome-wide association studies (GWAS), whole-exome sequencing (WES), and transcriptomic studies. Certain platforms organise these input genes into an output where genes are grouped according to biological processes/pathways. Tools like Gene Ontology (GO) highlights biological processes that may underlie the observed signal, the overrepresentation analysis of these biological processes is widely used to interpret gene lists from genetic studies. Yet, results depend critically on the background (universe/reference list) against which enrichment is tested. The GO Consortium explicitly advises using as background “the list of all genes from which your smaller analysis list was selected”. In practice, however, this step is frequently overlooked and the default option to use all genes in the genome as the background is selected. Wijesooriya et al.(2022) surveyed 186 papers and found that 95% of GO overrepresentation analysis (ORA) analyses either used an inappropriate background or failed to report it [1], [2], [3]. This widespread neglect is not just a procedural detail, it introduces bias results. When the background is mis-specified, the null model of enrichment is wrong, leading to systematic inflation or deflation of p-values and misleading sets of “enriched” GO terms. We examine how genome–exome background mismatch alters GO Biological Process significance and induces annotation-driven bias.

Methods

Simulations across multiple input gene list sizes are run to assess enrichment for lists from an exome universe tested against a genome background (and vice versa). The effect of background choice on GO Biological Process enrichment is quantified by overrepresentation analysis. To verify how background selection influences real study outputs, we analysed Brain-related gene lists from Genome-Wide Association Studies and whole exome sequencing studies.

Results : Evaluating genome, exome and protein-coding transcriptome derived input lists against genome-background, exome and protein-coding transcriptome input lists generate more significant GO IDs than genome-derived input lists. Comparing these enrichment results shows opposing effects: some GO IDs become significant more often under the genome background (inflation relative to the exome background), whereas others become significant less often (deflation). Results demonstrate that an incorrect background can introduce both Type I errors (false positives) and Type II errors (false negatives), and also that the default option of using a genome background can systematically distort GO Biological Process enrichment for exome derived inputs. The verification of simulation with brain gene sets illustrate that incorrect background selection can both introduce additional significant processes and suppress processes that are significant under an assay-appropriate background, influencing the interpretability of GO Biological Process enrichment outputs.

Conclusions

To identify and summarise background-dependent effects in a reproducible way, we developed a command code that runs GO enrichment under genome and exome backgrounds for the same input list and partitions significant GO IDs into background-common and background-unique sets. Our results show that failing to use an assay-appropriate background in genome, exome and protein-coding transcriptome studies can both introduce background-specific “significant” processes and mask processes that are significant under the correct background.

References

  1. ​​Wijesooriya, K., Jadaan, S.A., Perera, K.L., Kaur, T., and Ziemann, M. (2022). Urgent need for consistent standards in functional enrichment analysis. PLoS Comput. Biol. 18, e1009935. https://doi.org/10.1371/journal.pcbi.1009935
  2. Reimand, J., Isserlin, R., Voisin, V., et al. (2019). Pathway enrichment analysis and visualization of omics data using g:Profiler, GSEA, Cytoscape and EnrichmentMap. Nat. Protoc. 14, 482–517. https://doi.org/10.1038/s41596-018-0103-9
  3. Mubeen, S., Tom Kodamullil, A., Hofmann-Apitius, M., and Domingo-Fernández, D. (2022). On the influence of several factors on pathway enrichment analysis. Brief. Bioinform. 23, bbac143. https://doi.org/10.1093/bib/bbac143