Rchemcpp: An R package for computing the similarity of molecules

The functionality of Chemcpp is provided in R, that is the computation of similarities between molecules by kernel functions.
The following kernels are implemented:

• the marginalized graph kernel between labeled graphs
  
• extensions of the marginalized kernel
  
• Tanimoto kernels
  
• graph kernels based on tree patterns
  
• kernels based on pharmacophores for 3D structure of molecules
  

Input are molecules provided in the SDF format (MOL format) and output is a numeric kernel matrix. This kernel matrix easily be used in R for clustering and prediction tasks, like the prediction of toxicity or biological activity of a given molecule.

---

Publication:

G. Klambauer, M. Wischenbart, M. Mahr, T. Unterthiner, A. Mayr, and S. Hochreiter (2015) Rchemcpp: a web service for structural analoging in ChEMBL, Drugbank and the Connectivity Map. Bioinformatics. Advance access: doi: 10.1093/bioinformatics/btv373

Technical Report:

• Rchemcpp-TechReport.pdf
  

Availability:

• Download the source package available from Bioconductor:
  Rchemcpp R package
  
• A web service for finding structural analogs using Rchemcpp is online:
  Finding analogs in ChEMBL,Drugbank and the Connectivity Map - Similarity Search with Rchemcpp
  

Datasets:
A benchmarking data sets can be downloaded below:

• mutag.sdf (0.4 MB)
  

© 2012 by Institute of Bioinformatics, Johannes Kepler University Linz, A-4040 Linz, Austria [view imprint]