Skip to main content

Table 1 Metrics used for dissimilarity measurements

From: Implementation of multiple-instance learning in drug activity prediction

Dissimilarity Measure

Definitiona

Soergel

b + c a + b + c

Dice

b + c 2 a + b + c

Manhattan

b + c a + b + c + d

Rogers-Tanimoto

2 × b + c a + d + 2 × b + c

  1. a Let P1 and P2 be two pharmacophore fingerprints, a be the count of bits which are set to 1 in both P1 and P2, b be the count of bits which are set to 1 in P1 but not in P2, c be the count of bits which are set to 1 in P2 but not in P1, and d be the count of bits which are set to 0 in both P1 and P2. So a is called the number of total matches, b and c are called the number of single matches, and d is called the number of no matches.
Back to article page

BMC Bioinformatics

ISSN: 1471-2105

Contact us