PMGen/AFfine_PANDORA_modifications.txt at master · soedinglab/PMGen · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
###### General Modifications on Alphafold Fine-tune and PANDORA
Source-AFfine: https://github.com/phbradley/alphafold_finetune
Source-PANDORA: https://github.com/X-lab-3D/PANDORA

# Changes Made to Integrate Pandora and Affine Environments

## Modifications from Pandora Environment:
- Downgraded `python` from 3.11 to 3.9.19.
- Updated `muscle` to retain version 5.1 but included build details.
- Updated `blast` from 2.10 to 2.10.1 with build details.
- Added specific versions for:
  - `numpy`: 1.22.4
  - `pip`: 24.3.1
  - `setuptools`: 75.8.0
  - `pandas`: 2.2.3
  - `python-dateutil`: 2.9.0.post0
  - `biopython`: 1.80 with build details.

## Modifications from Affine Environment:
- Updated `ml-collections` from 0.1.0 to 0.1.1.
- Upgraded `numpy` from 1.21.0 to 1.22.4.
- Upgraded `pandas` from 1.3.4 to 2.2.3.
- Downgraded `scipy` from 1.7.3 to 1.7.0.
- Added CUDA-specific builds for:
  - `tensorflow`: 2.11.0
  - `tensorflow-estimator`: 2.11.0.

## Modifications to Alphafold:
 Due to numpy version modifications:
'alphafold/model/config.py' : np.int --> np.int32
'alphafold/data/templates.py': np.object --> object
'alphafold/model/config.py': print('valid models: ',CONFIG_DIFFS.keys()) added.

## Modifications to AFfine Pipline:
'run_prediction.py' flag --num_recycles added.
'run_prediction.py' flag --shuffle_templates added.
'run_prediction.py' flag --num_templates added.
'run_prediction.py': outfile_prefix = args.outfile_prefix+'_'+targetl.targetid --> outfile_prefix = args.outfile_prefix+targetl.targetid
'run_prediction.py': os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2' added.
'run_prediction.py': warnings.filterwarnings('ignore', category=FutureWarning) added.
'run_prediction.py': warnings.filterwarnings('ignore', category=DeprecationWarning)

## Modifications to PANDORA
'PANDORA/PANDORA/pandora_modifs.py' and 'PANDORA/PANDORA/Modelling_functions' , added netmhcipan_path, netmhciipan_path paths
'PANDORA/PANDORA/Modelling_functions' minimum modifications in anchor prediction and saving anchors.
'PANDORA/PANDORA/Modelling_functions' keep not only max score template, but all and sort them based on scores.
'PANDORA/PANDORA/PMHC/PMHC.py' modified to handle  not only the top similar MHC allele, but also the lower ranks
    in case of no similar one by adding arg `Target(max_num_rank=20)` checks up to 20 similar cases one by one.
    If NetMHCpan could predict anchor for any, uses that anchor. If not, uses canonical anchors.