Abstract
Memory plays a crucial role in human cognitive processes and daily functioning. While memory has a genetic basis, identifying the specific genetic factors influencing memory performance has proven challenging. This challenge arises because memory is a complex trait, whose genetic architecture likely comprises an accumulation of many low-effect size common variants. Thus, study sample sizes are easily insufficient, leading to underpowered genetic association analyses. This limitation is especially pronounced in studies that prioritize deep phenotyping over broader recruitment, resulting in smaller cohorts. Given these limitations, important biological signals may remain undetected when relying solely on conventional genome-wide significance thresholds. However, even variants that fall below these thresholds can yield meaningful insights when analyzed in a broader biological context. Therefore, we propose that relevant biological information can still be extracted by employing a less stringent p-value threshold paired with elaborate variant mapping and functional annotation to identify candidate variants, genes, and biological pathways associated with memory performance. We present three independent genome-wide association studies within the Human Connectome Project on a (i) Penn-Word verbal episodic memory test (n = 1131), a (ii) Picture Sequence visual episodic memory test (n = 1133), and a (iii) List Sorting verbal working memory test (n = 1134). Subsequent variant mapping, functional annotation, and pathway identification were performed using FUMA, Cytoscape, KEGG, Reactome, and WikiPathways. At the pathway level, but not single variant or gene level, we observed substantial overlap in results across the three memory tests, and between our results and previously reported findings. Several identified genes and pathways were previously associated with memory-related disorders, and processes related to cognition, neurodevelopment, and neurological dysfunction. We interpret the common pathways as reflecting shared biological mechanisms underlying memory. Our findings underscore the potential of our proposed approach, which we provide as an openly accessible pipeline, for exploring other complex polygenic traits.