Cluster Algorithms for Drug Discovery

Github : https://github.com/danielSoler93/cluster_drug_discovery

Cluster drug discovery is a specific python package to handle clustering in drug discovery. The main aim of the package is not to give a general cluster method that works in all cases (as it does not exist) but to give the possibility to clusterize in an easy manner and to asses the quality of the results.

Structure of the Package

The package consist of three parts:

1. An input preprocessing to rapidly convert features from your pdb, xtc, dcd to numerical values
2. The Clustering main class containing a high level object to wrap all cluster algorithms
3. An analysis package to asses how good is your cluster and what could be improved

Documentation

• Installation
  • From Conda (recommended)
  • From PyPi
  • From Source Code

• Getting Started
  • From python API
    • Analyze Dataset
      • Silhouette
      • Carlinski
      • David-Bouldin
      • Umap Projection
      • Analysis Conclusion
    • Clustering Algorithms
      • Kmeans
      • DBSCAN
      • Hierarchichal
  • From Command Line
    • Analyze your dataset
    • N most similar structures
    • Tanimoto similarity search
    • Search for one or more substructure
    • Search for at least one of the substructures in each sdf file
    • Search for similarity and substructure
    • Change fingerprint type
    • Use all fingerprints in one job with different tresholds
    • Turbo search method:

• ChangeLog
  • 1.0.0 - not yet there