pirl-unc / tsarina / 35374036341
79%

DEFAULT BRANCH: main
Ran
Jobs 1
Files 39
Run time 1min
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst
coverage: 76.961% (+0.2%) from 76.811%
35374036341

push

github

web-flow 
Make --cta's TPM optional, add default progress reporting and a table format (#162)

* Make --cta's TPM optional, add default progress reporting and a table format

Three usability fixes to `tsarina personalize`, found while walking
through a real patient run:

- --cta's "=TPM" half is now optional. A bare gene name (--cta SSX1,
  or {"SSX1": None}/{"SSX1": float("nan")} via the Python API) means
  "include this gene regardless of --min-cta-tpm", not "assume zero
  expression and gate it out."

- personalized_targets() now reports each stage (peptide generation
  per source, MS evidence lookup, presentation scoring) to stderr by
  default via a new show_progress parameter, wired through to
  cta_exclusive_peptides()'s existing on_progress/progress_bar
  support. Previously a multi-minute cold-cache run (generating every
  CTA candidate peptide, screening ~20K non-CTA genes for overlaps,
  then scoring against mhcflurry) produced no output at all until it
  finished or errored -- indistinguishable from a hang. --quiet
  disables it.

- New --format {table,csv} on the CLI (table by default when printing
  to the terminal, csv when writing to a file with --output, since CSV
  is for piping/loading elsewhere and a table is for reading). The new
  format_table() mirrors hitlist's pmhc --format table style (dashes
  under headers, no box-drawing) for a consistent look across the two
  tools' CLIs, though it's its own implementation since tsarina's
  one-row-per-peptide-with-a-chosen-best-allele schema doesn't map
  onto hitlist's per-(gene, allele) nested rows. A cta_flagged row's
  source gets a trailing '*' with a one-line footnote, rather than
  inlining the full flag_reason sentence into every such row.

Also fixed a real correctness bug surfaced by the same real-patient
run: aggregate_ms_hits_by_peptide's ms_alleles/ms_allele_count treated
each observation's whole raw mhc_restriction string as one token
before deduping, so a peptide seen across several... (continued)

3150 of 4093 relevant lines covered (76.96%)

0.77 hits per line

Coverage Regressions

Lines Coverage ∆ File
69
71.6
4.47% personalize.py
24
67.09
-4.34% cli_personalize.py
3
96.67
0.46% ms_evidence.py
Jobs
ID Job ID Ran Files Coverage
1 35374036341.1 39
76.96
GitHub Action Run
Source Files on build 35374036341