Abstract
Amyotrophic lateral sclerosis (ALS) is highly heritable, yet the vast majority of cases lack an identifiable genetic cause and clinical progression remains largely unpredictable. To connect noncoding and rare genetic variation to disease phenotypes in a relevant cell type, we generated a multi-omic quantitative trait locus (QTL) atlas from 594 induced-pluripotent-stem-cell-derived human motor neuron lines (522 ALS patients, 72 controls). By mapping cis-QTLs for chromatin accessibility, splicing and gene expression from whole-genome sequencing, we identify common and rare variants on the wild-type C9orf72 allele that form regulatory haplotypes. These haplotypes influence C9orf72 expression levels in motor neurons and stratify C9-ALS patients into four subgroups; using clinical disease duration data and longitudinal ALSFRS-R scores, we show that these subgroups exhibit different survival trajectories, indicating that wild-type C9orf72 expression acts as a genetic modifier of disease duration. Beyond the C9orf72 locus, we detect ultra-rare intronic variants that create cryptic exons and structural and nonsense variants in established ALS genes, providing likely genetic explanations for disease in additional patients who previously lacked a molecular diagnosis. Our results show that QTL mapping in patient-derived motor neurons can reveal regulatory modifiers of progression and hidden pathogenic events in ALS, providing a framework for genetically informed risk attribution and patient stratification in complex neurological diseases.