bayer-science-for-a-better-life / abbert-cerebras Goto Github PK

Hi,

Thanks for updating the code.
When I run the file examples/oas_consolidate_example.py I get the following error:

FileNotFoundError: Could not find the OAS root.
Please define the OAS_PATH environment variable or copy / link it to /Users/hossein/Desktop/Github/abbert-cerebras/data/oas-full

I think the error is because of line 8 in oas/ommon.py which defines:
RELATIVE_OAS_FULL_DATA_PATH = _RELATIVE_DATA_PATH / 'oas-full'

But we do not have the data for oas-full. If I change this line it to the following:
RELATIVE_OAS_FULL_DATA_PATH = _RELATIVE_DATA_PATH / 'oas-test'

The error goes away and I get the following in the output:

unit=('unpaired', 'Bhiman_2015', 'SRR2126755_Heavy_IGHM') chain=heavy ml_subset=train num_sequences=0
unit=('paired', 'Setliff_2019', 'SRR10313335_paired') chain=heavy ml_subset=test num_sequences=0
unit=('paired', 'Setliff_2019', 'SRR10313335_paired') chain=light ml_subset=test num_sequences=0

What is the purpose of oas_consolidate_example.py? I thought it should have given me one data frame as my dataset.
If I add unit.sequences_df(), it gives me three separate datasets with different sizes. Should I aggregate them? The first dataset has 41 columns but the other two have 76. I would appreciate it if you could clarify this.

Hi

When trying to run the file abbert-cerebras/examples/oas_data_examples.py, the following error is encountered:

  File "/cb/home/aarti/ws/code/bayer_external_repo/examples/oas_data_examples.py", line 4, in <module>
    from abbert2.oas import OAS_TEST_DATA_PATH, from_parquet
  File "/cb/home/aarti/miniconda3/envs/abbert2/lib/python3.9/site-packages/abbert2/oas.py", line 59, in <module>
    from antidoto.data import ANTIDOTO_PUBLIC_DATA_PATH
ModuleNotFoundError: No module named 'antidoto'

The installation procedure that was followed was

git clone [email protected]:bayer-science-for-a-better-life/abbert-cerebras.git
cd abbert-cerebras
conda env create -f environment.yml 

Could you also confirm if these are the correct installation steps ?

Thanks

For example, 'King_2020_2' -> 2. Should be easy to fix, or maybe we should just manually curate.

Hi,

I have noticed that if I do not comment line 318-332 in oas.py, I cannot run the code. It seems some records are missing in the dataset and that is why these exceptions are activated. If there were missing records why these exception handling were added?
If these exception handling was for sanity check of dataset, does this mean the dataset is incomplete or defective?

Filters like NoDuplicates and, more prominently, MergeRedundant, break the link between Paired VH/VL

How to deal with this? => Probably a many to many correspondence between "index_in_csv" of VHs and VLs needs to be kept around.

Not important unless we start training against paired VH & VL sequences (e.g., add a self-supervised goal to guess if a pair is real or not alongside training with both chains "concatenated")

As of 2021/11/14, we are missing 10 CSVs. Likely we need to report this to the OAS authors.

From oas diagnose:

http://opig.stats.ox.ac.uk/webapps/ngsdb/unpaired/Briney_2019/csv/SRR8283642_Heavy_Bulk.csv.gz
http://opig.stats.ox.ac.uk/webapps/ngsdb/unpaired/Briney_2019/csv/SRR8283642_Heavy_IGHA.csv.gz
http://opig.stats.ox.ac.uk/webapps/ngsdb/unpaired/Briney_2019/csv/SRR8283642_Heavy_IGHD.csv.gz
http://opig.stats.ox.ac.uk/webapps/ngsdb/unpaired/Briney_2019/csv/SRR8283642_Heavy_IGHE.csv.gz
http://opig.stats.ox.ac.uk/webapps/ngsdb/unpaired/Briney_2019/csv/SRR8283642_Heavy_IGHG.csv.gz
http://opig.stats.ox.ac.uk/webapps/ngsdb/unpaired/Briney_2019/csv/SRR8283642_Heavy_IGHM.csv.gz
http://opig.stats.ox.ac.uk/webapps/ngsdb/unpaired/Halliley_2015/csv/SRR1964711_Heavy_IGHE.csv.gz
http://opig.stats.ox.ac.uk/webapps/ngsdb/unpaired/Levin_2017/csv/ERR1812300_Heavy_IGHD.csv.gz
http://opig.stats.ox.ac.uk/webapps/ngsdb/unpaired/Thornqvist_2018/csv/ERR1812300_Heavy_IGHD.csv.gz
http://opig.stats.ox.ac.uk/webapps/ngsdb/unpaired/Tipton_2015/csv/SRR1964711_Heavy_IGHE.csv.gz