SniffCell Lite

SniffCell Lite keeps a compact workflow:

• sniffcell-lite find: call ctDMR catalogs from a methylation atlas.
• sniffcell-lite anno: annotate variants from supporting reads, a BAM, a reference FASTA, and a ctDMR catalog.
• sniffcell-lite report: build a lite-native HTML report, optionally with per-variant figures.

This repository is the canonical home of the lite find, anno, and reporting workflow.

Install

pip install sniffcell-lite

To install the latest development version directly from GitHub:

pip install git+https://github.com/Fu-Yilei/SniffCell-lite.git

Source code and issue tracking are maintained at Fu-Yilei/SniffCell-lite.

Find

find keeps the original atlas-driven flavor. The bundled tissue atlas supports tissue code or tissue name lookup through -ck.

sniffcell-lite find \
  -n atlas/all_celltypes_blocks.npy \
  -i atlas/all_celltypes_blocks.index.gz \
  -m atlas/all_celltypes.txt \
  -ck "Colon, Ascending" \
  -o colon_ascending.ctdmr.tsv

Equivalent code form:

sniffcell-lite find -ck 3E -o colon_ascending.ctdmr.tsv

The default cell-type JSON is the packaged sniffcell/data/tissue_atlas.json. You can still pass a custom atlas with -cf.

Anno

Single-variant mode requires a BAM and reference FASTA. SniffCell Lite maps the variant-supporting reads in the BAM, selects ctDMRs from the catalog that overlap those supporting-read alignment spans, computes methylation from the BAM only for those reads at those ctDMRs, and assigns the variant from the targeted read-level methylation calls. ctDMR evidence is not capped to a fixed distance from the variant.

sniffcell-lite anno \
  -i sample.bam \
  -r ref.fa \
  --variant-name variant_001 \
  --variant-location chr1:100000-101000 \
  --supporting-reads readA,readB,readC \
  --catalog colon_ascending.ctdmr.tsv \
  -o anno_out

Batch mode uses a TSV or CSV with these columns:

variant_name    variant_location    supporting_reads    catalog    bam    reference

Run:

sniffcell-lite anno --batch variants.tsv -o anno_out

supporting_reads accepts comma, pipe, semicolon, whitespace-delimited text, JSON list text, or @path/to/read_names.txt.

If every batch row uses the same reference, pass it once instead of adding a reference column:

sniffcell-lite anno --batch variants.tsv -r ref.fa -o anno_out

Outputs

sniffcell-lite find writes:

• *.tsv: ctDMR catalog
• *.tsv.igv.bed: IGV BED companion
• *.tsv.catalog.json: catalog manifest when tissue metadata was used

sniffcell-lite anno writes:

• variant_assignment.tsv
• variant_assignment_readable.tsv
• variant_assignment_readable_long.tsv
• reads_classification.tsv
• support_read_mappings.tsv
• anno_compact_manifest.json or anno_batch_manifest.json

Report

report reads a sniffcell-lite anno output folder directly. Batch reports keep the per-row BAM and ctDMR catalog from the original lite batch manifest, so multi-tissue/multi-BAM annotation outputs do not need to be coerced into a single full-SniffCell manifest.

Table-only report:

sniffcell-lite report --anno_output anno_out -o anno_report

Report with lightweight per-variant figures:

sniffcell-lite report --anno_output anno_out --with_figures -o anno_report

Useful filters:

sniffcell-lite report \
  --anno_output anno_out \
  --min_overlap_pct 0 \
  --min_majority_pct 0 \
  --max_variants 100 \
  --with_figures \
  -o anno_report

To report only listed variants, or to remove listed variants, provide a TSV/CSV with an id column or a one-ID-per-line file:

sniffcell-lite report \
  --anno_output anno_out \
  --include_variants selected_ids.tsv \
  --with_figures \
  -o selected_report

sniffcell-lite report \
  --anno_output anno_out \
  --exclude_variants rejected_ids.tsv \
  --with_figures \
  -o filtered_report

--include_variants and --exclude_variants are mutually exclusive. The report writes index.html, high_confidence_variants.tsv, report_manifest.json, normalized inclusion/exclusion provenance when used, and optional files under figures/.

Tests

pytest -q