PsRRR

PsRRR identifies gene pathways associated with multivariate quantitative traits by applying sparse reduced-rank regression with a group lasso penalty to genome-wide SNPs grouped into functional pathways for association with imaging endophenotypes.

Key Features:

• Sparse reduced-rank regression: Applies sparse reduced-rank regression to capture multivariate associations between genetic variation and quantitative traits.
• Group lasso on genome-wide SNPs: Implements a group lasso penalized regression framework modeling effects of genome-wide SNPs grouped into functional pathways based on prior gene-gene interactions.
• Pathway ranking via resampling: Ranks identified pathways using a resampling strategy that leverages finite sample variability to prioritize pathway importance.
• Integration with pathway databases: Integrates genome-wide SNP data with functional pathway information from databases such as KEGG.
• Voxel-wise longitudinal imaging analysis: Associates genetic variation with voxel-wise MR imaging signatures at multiple time points (6, 12, and 24 months relative to baseline).
• Application to whole genome scans and ADNI cohort: Has been applied to whole genome scans and MR images from the Alzheimer's Disease Neuroimaging Initiative (ADNI), including analyses of 99 probable AD patients and 164 healthy elderly controls.
• Gene and SNP prioritization: Facilitates investigation and prioritization of specific SNPs and genes driving pathway selection, highlighting PIK3R3, PIK3CG, PRKCA, PRKCB, ADCY2, ACTN1, ACACA, GNAI1, CR1, TOMM40, and APOE.

Scientific Applications:

• Alzheimer's disease research: Identifies pathways associated with AD endophenotypes, including insulin signaling, vascular smooth muscle contraction, and focal adhesion.
• Imaging genetics: Links genome-wide genetic variation to longitudinal structural brain changes measured by MR imaging endophenotypes.
• Gene and SNP discovery: Supports discovery and prioritization of candidate genes and SNPs implicated in AD biology, including genes linked to β-amyloid plaque formation and hippocampal expression changes.

Methodology:

Implements sparse reduced-rank regression with a group lasso penalty on genome-wide SNPs grouped into pathways using prior gene-gene interaction information; ranks pathways via a resampling strategy that leverages finite sample variability; integrates SNP data with KEGG pathway annotations; and analyzes voxel-wise MR imaging signatures at 6, 12, and 24 months relative to baseline.