Repeat and expand variant effect predictions

[adamklie]

Context

Expanding variant analysis across all model types and linking strategies.

Tasks

• Look for GWAS (LD expanded) intersections with peaks, then peak-to-gene links, then TF-to-peak links
• Score variants with ChromBPNet models trained on new pseudobulk peaks
• Score variants with CpGNet models trained on new pseudobulk methylation data
• Explore other scoring approaches (network-based?)

[adamklie]

Dependencies

• Blocked by Retrain ChromBPNet models at different resolutions #1 (ChromBPNet models), Finalize peak calling #5 (peaks for GWAS intersection), Retrain CpGNet methylation models on new pseudobulk data #14 (CpGNet models), Recall peak-to-gene links using Gaulton lab pipeline #2 (peak-to-gene links for intersection)
• Feeds into Prioritize cREs for CRISPRi validation #16 (cRE prioritization)

[adamklie]

Notes from project reorganization (2026-03-23)

Detailed TODO checklist filed in manuscripts/ChromBetaNet/bin/9_variant_effect_prediction/README.md:

T1D validated variants to test

From Chiou 2019: rs78037977, rs193778
From Gao 2022: rs35342456, rs193778, rs10483809

Analysis plan (from Chiou 2021 Fig 4)

1. SC-beta model T1D variant effect predictions → intersect with DARs → heatmap across models → examine attributions → call seqlets

Additional TODOs

• T2D variants
• Test validated variants with model
• Enrichment in peaks only?
• GWAS variant prioritization
• MPRA prediction accuracy
• caQTL prediction
• Intersect GWAS/finemap variants with SC-beta cCREs
• Figure 7: Variants in context of SC.beta GRN
• Check whether H1 genotype data exists
• Can SC-islet beta cell model predict mutation effects?

chr14 locus analysis

Detailed analysis of chr14:100,377,736-100,378,237 filed in updates/2026_02_01_chr14_analysis_report/README.md — includes peak calls, accessibility data, T1D credible set variants (DLK1 locus, ~347kb upstream), and cRE2Gene links to WDR25.

[adamklie]

Status Update (2026-04-09)

Blocked on #1, #2, #5, #14. Deferred.