Augsornworawat2023_sc-islet_10X-Multiome

10X Multiome (RNA+ATAC) of human stem-cell-derived islets across differentiation, transplantation, and CRISPRA/knockdown perturbations — comparator for our stimulated SC-islet multiome.

Source

• Reference: Augsornworawat P, Hogrebe NJ, Ishahak M, et al. Single-nucleus multi-omics of human stem cell-derived islets identifies deficiencies in lineage specification. Nat Cell Biol. 2023. TODO: confirm exact volume/page/doi from paper.
• Accession(s): SRA study SRP366403 (per results/1_get_data/SRP366403_metadata.tsv); GSM5979664+ series. TODO: corresponding GSE accession.
• Provenance: Downloaded from SRA/GEO via bin/1_get_data/{1_sra_download,2_geo_download}.ipynb, validated in 3_validate_downloads.ipynb.
• Local path: /cellar/users/aklie/data/datasets/Augsornworawat2023_sc-islet_10X-Multiome/ (symlinked from public-data/Augsornworawat2023_sc-islet_10X-Multiome/)

Study design

• Organism: Homo sapiens
• Assay: 10X Single Cell Multiome ATAC + Gene Expression v1 (paired snRNA + snATAC, same nuclei)
• Conditions / sample classes (from processed/ subdirs, 18 total):
  • SC-Islet differentiation timecourse: SC-Islet_Stage6_week_2, _week_3, _week_4, _month_6, _month_12
  • SC-Islet baseline replicates: SC-Islet_1, SC-Islet_2
  • SC-Islet perturbations: SC-Islet_ARID1B_knockdown, SC-Islet_GFP_Control
  • CRISPRA: CRISPRA_SC-Islet_CTCF_Doxycycline, CRISPRA_SC-Islet_CTCF_Untreated_Control
  • Transplanted SC-Islets: Transplanted_SC-Islet_{1,2,3}
  • Primary human islet controls: Human_Islet_{1,2,3,4}
• Run-level metadata: 40 SRR runs (paired ATAC + RNA libraries) in results/1_get_data/SRP366403_metadata.tsv.
• Genome build: hg38 + GENCODE (per CellRanger-ARC processing; project's TODAY.md notes "hg38+GENCODE full compat").
• Sample sheet: results/1_get_data/sample_metadata.tsv lists only 7 SC-Islet samples — partial coverage of the 18 processed dirs. TODO: reconcile sample sheet with full processed sample set.

What's here

bin/
  1_get_data/         SRA + GEO download notebooks, scripts/
  2_process_data/     CellRanger-ARC counting (count_reads, prep, SLURM array)
  3_sample_qc/        RNA QC, ATAC QC, AMULET doublet detection, configs/
  4_cell_annotation/  prep, merge_RNA, analysis_RNA, integrate_RNA, cell_annotation, merge_ATAC
  TODO.md             generic template stub (not dataset-specific notes)
processed/            18 per-sample CellRanger-ARC output dirs (outs/, .mri.tgz, etc.)
results/
  1_get_data/         SRP366403_metadata.tsv, sample_metadata.tsv, fastq_stats.tsv, srr_ids.txt
  2_process_data/     per-sample CellRanger-ARC summary CSVs (10 samples present)
  3_sample_qc/        per-sample QC outputs (rna/) — 7 SC-Islet samples present
  4_cell_annotation/  rna/{merge, normalize, reduce_dimensions, integrate/round_1, analysis}
                       analysis/{cell_cycle_scoring, clusters, embeddings, marker_genes, pathways}
                       1_integrate_RNA.tsv (integration manifest)
                       NOT present: atac/ outputs (ATAC integration not yet run)
scratch/              exploration
env/                  conda env spec
README.md             (this file; previous root README was a generic template — see bin/README.md)

Status

• Pipeline state: annotated (RNA only); ATAC processed per-sample but not integrated.
• Latest run: TODO: confirm date from results/4_cell_annotation/rna/ mtimes.
• Current limitation / blocker: RNA fully processed (~44k cells, annotated, hg38+GENCODE compat) but ATAC integration is the gap — fragments/peaks exist per sample, no unified ATAC h5ad. bin/4_cell_annotation/6_merge_ATAC.sh exists but no merged ATAC outputs in results/.
• Owner: aklie

Use in this project

• Why we have it: External SC-islet multiome benchmark — independent differentiation protocol and perturbation panel against which to compare our stimulated SC-islet annotations, gene programs, and chromatin features.
• Linked issues: IGVF-UCSD/.github#9 (Augsornworawat integration).
• Scratch workspace: TODO: confirm path under igvf-data/igvf_sc-islet_10X-Multiome/scratch/YYYY_MM_DD/.
• Manuscript: beta_cell_networks