Follow the changelog format from https://keepachangelog.com/en/1.0.0/.
For later changes, see the GitHub releases.
- Support transfer learning based on DIA search results.
- Add
peptdeep cmd-flowCLI command for CLI users, see README for details. - Add
ThermoTOFandAstralinstrument type for Astral
other_modification_mappingtopsm_modification_mappingin yaml.- Removed
rescorein CLI and removedpercolatorin global_settings.
- Enable
speclib_tsvfor transfer learning in CLI and GUI.
- Fatal: transfer learning failed in
match_psms()with error"... match_psms() takes 0 arguments but 1 were given ..."in GUI and CLI.
- 'protein_reverse' decoy, see https://alphabase.readthedocs.io/en/latest/protein/protein_level_decoy.html
- Enabled multi-target prediction, all targets will store in one column as np.array object
- Enabled
attention_maskand fixed sequence length - Tested
torch.jit.traceto compile the JIT model
- Multiprocessing with
spawnfor different OS systems to prevent hangs on Linux. - Using constant values as defaults params of class methods or functions instead of
global_settings[xxx]values. Usingglobal_settings[xxx]values as defaults params does not update onceglobal_settingschanges, this is dangerous.
- Testing on Python 3.10
fixed_sequence_lenwith padding zeros for sequence models.- GUI: it is able to delete tasks in the task queue.
user_defined_modificationsinpeptdeep/constants/global_settings.yamlfor CLI and GUI, it allows us to define our own modifications.other_modification_mappinginpeptdeep/constants/global_settings.yamlfor CLI and GUI, it allows us to read arbitrary PTMs from other search engines.special_modsin global_settings for special modifications like GlyGly@K or Phospho.labeling_channelsin global_settings.- MS2/RT/CCS models for Dimethyl-labeled peptides, see https://github.com/MannLabs/alphapeptdeep/releases/tag/dimethyl-models.
- "No match found for given type params" for
IntPtr.__overloads__[Int64]inpythonnet>=3forDotNetArrayToNPArray(), see https://github.com/MannLabs/alphapeptdeep/blob/main/peptdeep/legacy/thermo_raw/pyrawfilereader.py#L77
- Test the model after transfer learning (MS2/RT/CCS), and the metrics will be logged.
frag_end_idx->frag_stop_idx
- More features (mostly about modifications) in GUI.
- Peptide labeling and special modifications in
peptdeep.protein.fasta.PredictSpecLibFasta. - Use queue system in
peptdeep/webui/serve.pyto reduce conflicts of prediction.
- Use sphinx and readthedocs for documentation, nbdev_test for unit tests.
- Plan to move thermo_raw to AlphaRaw.
- First official release
- FIXED in GUI: keep_only_important_modloss -> modloss_importance_level
- First release
- FEAT: Initial creation of peptdeep