Prerequisites: Ubuntu 16.04, Anaconda from https://www.anaconda.com/download/#linux, Chrome Browser (for jupyter notebook)
Run:
./bootstrap.py
source activate cs510-env
jupyter notebook
Then open classifier.ipynb in the opened notebook
Notes:
To get compounds used in assay 1032, do
cat /media/data/pubchem/Data/0001001_0002000/1032.csv | cut -d, -f3 > CIDS_ASSAY_1032.csv
This file can be uploaded to (PubChem)[https://pubchem.ncbi.nlm.nih.gov/pc_fetch/pc_fetch.cgi]
TODO:
Explore graph conv and word2vec derived fingerprints
Graph conv: https://arxiv.org/abs/1603.00856
mol2vec fingerprints: https://chemrxiv.org/articles/Mol2vec_Unsupervised_Machine_Learning_Approach_with_Chemical_Intuition/5513581