Skip to contents

scphasedcn 0.1.1

  • Extractor (Python): the read_id of a read whose name does not follow the SPLONGGET convention (<CB>_#<uuid>_<sign><piece>of<npieces>) is now a stable 64-bit blake2b digest of the name. It was Python’s hash(), which is salted per process, so for BAMs from other pipelines the read_id column of fragments.tsv.gz changed from run to run. Reads named in the SPLONGGET convention are unaffected, and so are depth, alleles and the caller.
  • No changes to the R package.

scphasedcn 0.1.0

First public release of scphasedcn.

  • call_scn(): copy-number calls per cell and region from an extraction directory (scphasedcn run), combining read depth and read-phased alleles.
    • Normal-tissue mode (caller_normal()): null cells selected from the data, Student-t / beta-binomial nulls, fold-cross-validated thresholds at 1 false call per 1,000 null cell-regions, per-cell and pooled calls, events.
    • Tumour mode (caller_tumour()): clone discovery, allele-specific clone copy-number states, likelihood assignment of every cell, per-cell inherited and confirmed calls, MEDICC2 input (medicc2_pieces(), medicc2_table()).
  • write_scphasedcn() / read_scphasedcn(): flat result tables, the parameters used (params.yaml) and run notes.
  • Command line: system.file("scripts", "call_scn.R", package = "scphasedcn"), on --dir + --out, or on a sample yaml (inst/extdata/sample_template.yaml) with --config + --sample.
  • make_cell_meta(): the cell metadata table (cell types, known copy-number-flat cells, groups) from the extractor’s cells_qc.tsv and your own labels.
  • The hg38 centromere table ships with the package (hg38_centromeres()); the chrX centromere is masked like the autosomal ones.
  • simulate_scn() / write_sim_dataset(): simulated SPLONGGET-like data in the extractor’s layout.
  • Vignettes: getting started, inputs (phasing recipe, extraction, cell metadata), outputs.