# README main/
##########################################


# abstraqt.pkl
##########################################

Contains the Ball Tree used for the kNN regressor
Unpacks into a 4-uple:
- trees (list of Ball Trees (sklearn objects), one per distance cutoff)
- labels (list of lists of binaries, one per distance cutoff per training point)
- cutoffs
- percentages (for the selected Arity percentiles. Default: (25, 75))

Usage:
sbl-alphafold-abstraqt.py --mode scoring --weights abstraqt.pkl -f your_structure.pdb


# create_db_scripts/
##########################################

Contains all scripts to generate the training/validation database for ABSTRAQT

## Setup
Before creating the database, you must:
- Download the PDB structures listed in the file
- Download the whole AFDB human proteome
- Run AIUPred on all the structures and store the results in one text file
- Edit the create_db_scripts/locations.sh 


## create_db.sh
Main script



## check_requirements.sh create_db.sh                  generate_equilibrated_dataset.sh  requirements.sh
compare-af2rsa-disprot.py  find-unconfirmed-disorder.py  report_unphysical_SSEs.py