Predicting kinase inhibitors using bioactivity matrix derived informer sets

This repository contains baseline Informer-Based Ranking (IBR) methods and procedures for evaluating metrics for IBR performance (including non-baseline IBRs). It provides the kinase screening data used to evaluate the IBR methods. See also the repositories for:

• Regression Selection (RS)
• Coding Selection (CS) and Adaptive Selection (AS)

Citation

If you use this software or the new chemical screening data, please cite:

Huikun Zhang⁺, Spencer S Ericksen⁺, Ching-pei Lee⁺, Gene E Ananiev, Nathan Wlodarchak, Peng Yu, Julie C Mitchell, Anthony Gitter, Stephen J Wright, F Michael Hoffmann, Scott A Wildman, Michael A Newton. "Predicting kinase inhibitors using bioactivity matrix derived informer sets." PLoS Computational Biology 15:8, 2019.

⁺ equal contributions

File Structure

The informers repository comprises 5 main folders:

• source - code for running baseline methods and running evaluation metrics for all IBRs (both validation and new targets)

• data - new screening data for PknB and BGLF4, pre-processed PKIS1 and PKIS2 screening data

  • data/compounds - compound SMILES, Morgan fingerprints, and Morgan Jaccard distance matrices
  • data/thresholds_2sigma - inferred target activity thresholds for assigning compound binary activity labels
  • data/original_data - original PKIS1 and PKIS2 data sets with descriptions of pre-processing
  • data/rop18 - PKIS1 activity data from assays on Toxoplasma gondii Rhoptry Kinase ROP18, Simpson et al. 2016

• output_newtargs - output from all IBR methods on prospective microbial kinase targets (PknB, BGLF, ROP18) and metrics evaluations

• output_pkis1loto - output from all IBR methods for 224 PKIS1 targets and metrics evaluations

• figures - codes for plotting figures

Python environment

This code was run in the following conda environment:

Python version: 2.7.15

Packages in environment:

Name	Version
matplotlib	2.2.2
numpy	1.14.5
pandas	0.23.3
rdkit	2018.03.2
scikit-learn	0.19.2
scipy	1.1.0
seaborn	0.9.0