Abstract B007: Functional and epigenetic characterization of African pangenome contigs: Implications for reference genome bias and human genomic diversity

Despite significant advancements in genome assembly, widely used human references, such as GRCh38 or the complete models like T2T-CHM13, still exhibit limitations in capturing the extensive genomic diversity present within African populations. In this study, we investigated 124,240 African Pan-Genome (APG) contigs that were previously unaligned to GRCh38 for their functional and structural characteristics. Using BWA-MEM, we aligned these contigs to T2T-CHM13 and 47 linear assemblies from the Human Pangenome Reference Consortium (HPRC) at both stringent (nearly perfect) and relaxed (reasonably good) alignment thresholds. RepeatMasker was utilized to assess the repeat composition of contig subsets, and ab initio protein-coding gene and CpG island (CGI) predictions were performed using AUGUSTUS and EMBOSS CpGplot. At the nearly perfect threshold, 40% of APG contigs aligned to T2T-CHM13, while 83% aligned to HPRC. Notably, 94.5% of contigs aligned to T2T-CHM13 were in regions unique to that assembly compared to GRCh38, including centromeres and satellite repeats. Overrepresentation analysis revealed significant enrichment of contigs linked to genes associated with biological processes related to cellular immunity, synaptic function, and intracellular signalling. A population-level alignment summary across the HPRC assemblies indicated distinct ancestry-related enrichment in African and Admixed American genomes, suggesting the unique structural and evolutionary characteristics of African genomes. Out of 21,187 APG contigs that did not map using the strict alignment threshold, 14,796 aligned to both references under relaxed criteria, and 5,649 aligned exclusively to HPRC. The remaining 742, referred to as below-threshold contigs, either failed to align or aligned at below the relaxed threshold. The aligned contig groups had over 93% repeat content, primarily comprising satellite and simple repeats, whereas the subset below the threshold averaged only 19% repeat content. Functional annotation identified multiple predicted protein-coding genes and CGIs within the below-threshold contigs. Specifically, 58.5% contained only predicted genes, 0.3% contained only CGIs, 3.6% included both, and 37.6% of contigs remained uncharacterized by these criteria. These results underscore the limitations of existing reference assemblies and highlight the need to incorporate diverse populations into reference genome catalogs, ultimately enhancing equity in genomic research and precision medicine. Citation Format: Rachel Martini, Abdulfatai Tijjani, Kyriaki Founta, Daniel Cha, Sebestian Maurice, Jason White, Onyinye Balogun, Melissa B. Davis, Nyasha Chambwe. Functional and epigenetic characterization of African pangenome contigs: Implications for reference genome bias and human genomic diversity [abstract]. In: Proceedings of the 18th AACR Conference on the Science of Cancer Health Disparities; 2025 Sep 18-21; Baltimore, MD. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2025;34(9 Suppl):Abstract nr B007.