v3.1.0
Added
- ✨ New
tims2rescore
CLI entrypoint with TIMS-specific configuration defaults - ✨ IM2Deep feature generator for ion mobility
- ✨ Support for Bruker raw (
.d
) and miniTDF formats (was also back-ported to v3.0.3) - ✨
psm_id_rt_pattern
andpsm_id_im_pattern
: New options to allow parsing of RT/IM from PSM file spectrum identifiers. - ✨
max_psm_rank_input
andmax_psm_rank_output
: New options to control rescoring and reporting of multiple PSMs per spectrum (default 5 and 1) - ✨ Also store peptide-level q-value and PEPs in the output PSM list, under the
metadata
field. - ✨
profile
option to write a cProfile report of the rescoring process (for debugging) - ✨ Option to pass
train_fdr
to Mokapot (test_fdr
was already supported as kwarg) - ⚡ Faster spectrum file reading with
timsrust
andmzdata
Rust libraries - 📝 Add TIMS²Rescore docs and citation on README; more detailed docs on formula modifications
Changed
- 👷 CI: Lint with ruff instead of flake8
- ⚡ DeepLC: By default use PSMs initially passing 1% FDR threshold for calibration instead of a fixed number
- 🔊 DeepLC: Throw warning when using
deeplc_retrain
with less than 500 calibration PSMs - 🔊 Raise exception if missing precursor info (retention time or ion mobility) could not found in both PSM and spectrum files.
- ⬆️ Update (and sort) dependencies in
pyproject.toml
- 📦 Dockerfile: Use Python 3.11 base image instead of Ubuntu
- 🔧 Config: Change
write_report
default value fromtrue
tofalse
Fixed
- 🐛 Report: Fix calculation of number of retained peptides in identification overlap chart
- 🐛 DeepLC: Fix issue with conversion to PEPREC (fixes #108, fixes #113)
- 🐛 Fix bug in infer_spectrum_path if path in configuration is None
- 🐛 Fix bug that always set
im_required
toTrue
, even if no IM-based feature generators were configured.
Full Changelog: v3.0.3...v3.1.0