Flat tables: cells.tsv (cell, qc, stratum, fold, null_cell, n_calls, n_events, clone / posterior /
doublet_like / ref_pool when present), calls.tsv (cell- and clone-level calls: called rows in normal mode,
all inherited / confirmed rows in tumour mode), pooled_calls.tsv (called pool rows), events.tsv,
thresholds.tsv, null_fit.tsv (depth t rows + rho), false_calls.tsv + false_events.tsv + expected_false.tsv
(normal mode; event_fdr()), null_selection_check.tsv (normal mode with call_scn(null_check = TRUE)),
tumour mode: clones.tsv, assignments.tsv, clone_segments.tsv, clone_profiles.tsv.gz,
reference_pool.txt; regions.tsv; result.rds (the object without scores and features), params.yaml,
summary.tsv (scphasedcn_summary()) and NOTES.md. Optional: scores.tsv.gz, features.tsv.gz.
The files of a previous result in out (the names above) are removed first, so a re-run never leaves stale
tables behind (e.g. clones.tsv of a tumour run, scores.tsv.gz of a run with write_scores); other files in
out are left alone. An object without scores (or features), e.g. one read back with read_scphasedcn(),
leaves an existing scores.tsv.gz (features.tsv.gz) in place: it cannot write them again.
Arguments
- obj
scphasedcnfromcall_scn()- out
output directory (created)
- write_scores
also write the per cell x region scores
- write_features
also write the features long table