Despite advancements in short-read sequencing (SRS), over 50% of individuals with a suspected monogenic condition lack a genetic diagnosis after undergoing clinical testing. In addition to interpretative challenges, one plausible reason is that pathogenic variants reside in parts of the genome that are refractory to SRS techniques. Long-read sequencing (LRS) offers significant advantages over SRS, including superior detection of structural variants, better resolution of repetitive regions, accurate long-range haplotype phasing, and direct DNA methylation detection. However, the clinical adoption of LRS is hindered by cost and the unquantified potential for conclusive diagnosis. To overcome these challenges, I propose using cost-effective LRS protocols and informatics to improve diagnostics in rare diseases.

In my first aim, I evaluate the effectiveness of cost-efficient long-read nanopore sequencing by applying NAPU to a cohort of rare disease cases that previously yielded inconclusive results using SRS. Given that LRS enables simultaneous analysis of both genetic variants and methylation patterns, my second aim is to develop scalable and statistically robust methods for performing differential methylation analysis in both the rare disease and CARD cohorts. With over 4,000 genomes from patients with neurodegenerative disorders and control samples, the CARD dataset offers a unique opportunity to establish baseline methylation patterns and explore their associations with ancestry, age, sex, and disease status. Accurately characterizing variation in medically relevant genes within segmental duplications remains a challenge, even with LRS, due to the complexities of homology. Traditional alignment and variant calling tools often struggle in these regions with paralogous gene families, where crossovers obscure differentiation between gene copies. Therefore, in my third aim, I propose to expand Parakit, a pilot tool that utilizes collapsed pangenomes to detect pathogenic variants and predict haplotypes in the CYP21A2/CYP21A1P gene family. I will enhance this tool to automate its application across any paralogous gene family and subsequently perform population-wide haplotype analysis to identify haplogroups and estimate their population allele frequencies.

Event Host: Shloka Negi, Ph.D. Student, Biomolecular Engineering & Bioinformatics

Advisors: Dr. Benedict Paten and Dr. Karen Miga