Skip to content

Releases: weizhenFrank/DeepPhospho

20231211

11 Dec 14:27
Compare
Choose a tag to compare

Changes in this release:

  • Add arguments to control the minimum and maximum numbers of "fragments per precursor", and minimum relative fragment intensity, in library generation step (both runner and app are modified) (2ef0ee6)

Files in this release:

  • DeepPhospho-20231211.zip has updated code but no enviroment included
  • DeepPhospho-20231211-WinENV.zip has both updated code and enviroment included

If you have downloaded DeepPhospho with an python enviroment included from release page before, just download the smaller one (DeepPhospho-20231211.zip) and copy the folder DeepPhosphoPythonEnv from old version to main folder of new version will save much time

20230528

28 May 12:49
Compare
Choose a tag to compare

Changes in this release:

  1. Numpy with version >= 1.20 now works (0002a6b)
  2. Help message for RT model of pretrained weight files and skip fine-tuning is simplified (02554b4)
  3. Use library generated from FragPipe as training data now works in runner (2d5c5e5)

Files in this release:

  • DeepPhospho-20230528.zip has updated code but no enviroment included
  • DeepPhospho-20230528-WinENV.zip has both updated code and enviroment included

If you have downloaded DeepPhospho with an python enviroment included from release page before, just download the smaller one (DeepPhospho-20230528.zip) and copy the folder DeepPhosphoPythonEnv from old version to main folder of new version will save much time

Please note: current restrictions for a generated library are

  • fragments per precursor 3-15
  • min fragment relative intensity 5%
  • fragment series number >=3
  • fragment vs precursor m/z diff > 0.055

20221026

26 Oct 09:54
Compare
Choose a tag to compare
  1. Set number of workers of all data loaders to 0 (0587c0b)
  2. Fontsize and window size can be changed in GUI (File -> Set size) (8c496d9)
  3. Add demo for ion mobility training and prediction, based on pre-trained ion intensity model (now could only run with scripts /ion_mobility_demo/tuning_on_ion_pretrain.py and /ion_mobility_demo/im_pred.py) (a779fd9)

Files in this release:

  • DeepPhospho-20221026.zip has updated code but no enviroment included
  • DeepPhospho-20221026-WinENV.zip has both updated code and enviroment included

If you have downloaded DeepPhospho with an enviroment included from release page before, just download the smaller one (DeepPhospho-20221026.zip) and copy unzipped files to the folder in old version will save much time

20211212

12 Dec 11:15
Compare
Choose a tag to compare
  1. Add runner argument -tr / --train_split_ratio to control split ratio of training data
  2. Set default ratio from 9:1 to 8:2
  3. Support training with tsv file generated by EasyPQP. This needs column "Annotation" correctly existed
  4. Support prediction with tsv file from EasyPQP, and peptide list with UniMod format
  5. Add pack to zip arg to desktop packing script
  6. Fix used column name of MQ msms file as training data
  7. Re-arrange evaluation part for pearson and delta95 calc in rt pred function to avoid error with single input (now need at least 3 to trigger evaluation report)
  8. Add first char check for DP format. If both acetyl symbol '@' and '*' are not defined, will have '@' (no acetyl mod) as default
  9. Check if peptide contains invalid char at preprocessing step (will skip them but not raise error now)
  10. Change default epoches for ion and RT model training from 30 to 20 in ui
  11. Change number of kept fragments from 30 to 15 and number of min required fragments per prec from 3 to 4 in library generation

Files in this release:

  • DeepPhospho-20211212.zip has updated code but no enviroment included
  • DeepPhospho-20211212-WinENV.zip has both updated code and enviroment included

If you have downloaded DeepPhospho with an enviroment included from release page before, just download the smaller one (DeepPhospho-20211212.zip) and copy unzipped files to the folder with old version will save much time

V1.0.0

24 Oct 03:05
Compare
Choose a tag to compare
Update download link for pre-trained weights

20210719 packed source code and python env

20 Jul 04:54
Compare
Choose a tag to compare