Personalized and graph genomes reveal missing signal in epigenomic data

Area of Science:

• Genomics
• Epigenetics
• Bioinformatics

Background:

• Next-generation sequencing epigenomic studies commonly align reads to a reference genome.
• Genetic diversity and the diploid human genome can cause misalignments and bias results when using generic references.

Purpose of the Study:

• To investigate the impact of genetic variation on epigenomic data analysis.
• To evaluate the effectiveness of personalized and graph genomes compared to generic references.

Main Methods:

• Comparison of histone H3K4me1 and H3K27ac ChIP-seq peak calls using generic, modified, and de novo assembled reference genomes.
• Analysis of read distribution and variant impact (indels, SNVs) on peak alterations.
• Evaluation of graph personalized genomes as a compromise approach.

Main Results:

• Modified and de novo genomes alter ChIP-seq peak calls, creating new peaks or losing reference peaks.
• De novo genomes altered up to 5% of peaks, while modified genomes altered ~1%.
• Short insertions/deletions (indels) and single nucleotide variants (SNVs) most frequently modified peak calls.

Conclusions:

• Personalized and graph genomes enable the discovery of new epigenomic peaks influenced by indels and SNVs.
• These altered peaks exhibit inter-individual variability and may be relevant for studying human phenotypes.