Reprocess snm3C-seq with new samples

[adamklie]

Context

For igvf-data/igvf_sc-islet_snm3c, reprocess with new samples.

Tasks

• Transfer new data from TSCC to cluster
• Run through pipeline (bin/1_process) and sample QC (bin/2_sample_qc)
• Deal with batch correction
• Continue downstream analysis steps

[adamklie]

Dependencies

• Feeds into Integrative analysis: snm3C-seq × Dominguez2020 bulk methylation #12 (snm3C × Dominguez2020 integration), Retrain CpGNet methylation models on new pseudobulk data #14 (CpGNet retraining)

[adamklie]

Notes from project reorganization (2026-03-23)

Detailed processing and integration documentation filed in igvf-data/igvf_sc-islet_snm3c/README.md:

Data integration approach

• Modality preprocessing: snmCG gene body methylation (55,486 features), snRNA, snATAC 5kb bins, snATAC gene activity, snmCG 5kb hypomethylation
• Joint embedding via CCA across modalities
• Consensus cell type annotation by majority label across 3 integration strategies

scHiCluster pipeline status

Imputation, gene scoring, compartments (single-cell), pseudobulking, loop calling, and embedding are done. Domains failed (R issue). Differential analysis still TODO.

Outstanding

• Differential methylation: pseudobulk DMG (t-test + ALLCools), single-cell DMG, DMRs (MethSCAN, methylpy)
• Figure 5 plans: differential loops (5a), compartments (5b), boundaries (5c), gene scores (5d)
• mCG pseudobulking for celltype+treatment: ALLC + bigwig generation done

[adamklie]

Status Update (2026-04-09)

Unblocked but deferred — this has many steps and we're prioritizing multiome analyses (#1, #4, #8) first. Will come back to this.