Information theoretic clustering of the human pangenome minigraph

Information theoretic clustering, long-range correlation, power-law scaling and self-similarity concepts have been broadly adopted for characterizing genomic features such as nucleotide composition, flexibility and bending. In this work, the 24 chromosomes of the human pangenome minigraphs, recently assembled by the Human Pangenome Reference Consortium (HPRC), are investigated to check to what extent self-similarity and scaling features are preserved in comparison to the reference linear sequences of the T2T-CHM13 individual. By taking the nucleotide self-similarity of the reference chromosomes as benchmark, it is shown that the pangenome minigraph segments exhibit lower self-similarity of the nucleotide composition compared to the linear sequence. The proposed information measures can be adopted to quantify the nucleotide self-similarity patterns and complement standard alignment techniques towards the coherent definition of the genomic profile of each species. • Genomic string of text (A, G, C, T chars) are mapped to numerical sequences. • Order/disorder of nucleotides within the string is relevant to biological processes. • Clustering of nucleotide strings provide insights on genomic internal organization. • Pangenome minigraphs gather together information content from same species sequences. • Clustering of minigraph quantifies compositional order of human pangenome.