iSAR

iSAR

Perl implementation of a lazy SAR algorithm
- feature generation: subgraphs and paths with gSpan or unrooted trees with Free Tree Miner
- k nearest neighbor classification
- local models

- validation: LOOCV
- classification is tailored for test instance
- derive test compounds substructures and use them to determine similar structures
- possibility to use also strongly activating and deactivating fragments from training set
- possibility to use redundancy removement of fragments (use only closed features)
- KNN classification for one k or several values of k possible (voting among resulting predictions)

More detail:
http://wwwkramer.in.tum.de/research/pubs/articlereference.2008-03-17.2708343675

• Subproject of: OpenTox
• Data: No
• Software: Yes
• Programming language: perl
• Operating system: linux
• Dependencies: Weka, JOELib, FTM or gSpan'
• Interface: command line
• Inter-process communication: File IO
• Input format: lazySAR internal format (.ibf, .fbi, .count)
• Output format: .result, .info (plain text files)