Skip to content

FPSim2's documentation

FPSim2 is a NumPy-centric Python/C++ package for running fast compound similarity searches. FPSim2 performs optimally with high search thresholds (>=0.7). It is currently used in the ChEMBL and SureChEMBL interfaces.

Highlights

  • Uses SIMD instructions
  • Implements bounds for sublinear speedups from 10.1021/ci600358f
  • Features a compressed file format with optimized read speed based on PyTables and BLOSC
  • Provides fast multicore CPU and GPU similarity searches
  • Supports in-memory and on-disk search modes
  • Enables distance matrix calculation

Getting started

Install and generate a FPSim2 fingerprint db file.

CPU searches

Learn how to run CPU based similarity searches:

GPU searches

Learn how to run GPU based similarity searches: