Skip to main content

Table 3 Average number of fragments and bit-collisions when folding circular fingerprints on our datasets

From: Filtered circular fingerprints improve either prediction or runtime performance while retaining interpretability

Type Fragments 1024 2048 4096 8192
Rate Bit-load Rate Bit-load Rate Bit-load Rate Bit-load
ecfp6 80,342.54 1 78.46 0.99 39.24 0.98 19.64 0.95 9.86
ecfp4 23,874.58 0.99 23.32 0.98 11.68 0.94 5.89 0.8 3.11
ecfp2 2169.37 0.7 2.39       
ecfp0 57.01         
  1. Rate is the ratio of bit positions that are mapped by more than one fragment (e.g., 99% of bit-positions correspond to multiple fragments for ECFP4 and bit-vector size 1024). Bit-load is the mean number of fragments that are mapped to a single bit