/fingerprint/man/distance.Rd

  1\name{distance}
  2\alias{distance}
  3\title{
  4Calculates the Similarity or Dissimilarity Between Two Fingerprints
  5}
  6\description{
  7  A number of distance metrics can be calculated for binary
  8  fingerprints. Some of these are actually similarity metrics and
  9  thus represent the reverse of a distance metric.
 10
 11  The following are distance (dissimilarity) metrics
 12  \itemize{
 13    \item Hamming
 14    \item Mean Hamming
 15    \item Soergel
 16    \item Pattern Difference
 17    \item Variance
 18    \item Size
 19    \item Shape
 20  }
 21
 22  The following metrics are similarity metrics and so the distance can
 23  be obtained by subtracting the value fom 1.0
 24  \itemize{
 25    \item Tanimoto
 26    \item Dice
 27    \item Modified Tanimoto
 28    \item Simple
 29    \item Jaccard
 30    \item Russel-Rao
 31    \item Rodgers Tanimoto
 32    \item Cosine
 33    \item Achiai
 34    \item Carbo
 35    \item Baroniurbanibuser
 36    \item Kulczynski2
 37    \item Robust
 38  } 
 39
 40  Finally the method also provides a set of composite and asymmetric
 41  distance metrics
 42  \itemize{
 43    \item Hamann
 44    \item Yule
 45    \item Pearson
 46    \item Dispersion
 47    \item McConnaughey
 48    \item Stiles
 49    \item Simpson
 50    \item Petke
 51  }
 52  The default metric is the Tanimoto coefficient.
 53}
 54\usage{
 55distance(fp1, fp2, method)
 56}
 57\arguments{
 58  \item{fp1}{
 59    An object of class \code{fingerprint} or \code{featvec}
 60  }
 61  \item{fp2}{
 62    An object of class \code{fingerprint} or \code{featvec}
 63  }
 64  \item{method}{
 65    The type of distance metric desired. Partial matching is
 66    supported and the deault is \code{tanimoto}. Alternative values are
 67    \itemize{
 68      \item \code{euclidean} 
 69      \item \code{hamming}
 70      \item \code{meanHamming}
 71      \item \code{soergel}
 72      \item \code{patternDifference}
 73      \item \code{variance}
 74      \item \code{size}
 75      \item \code{shape}
 76
 77      \item \code{jaccard}
 78      \item \code{dice}
 79      \item \code{mt}
 80      \item \code{simple}
 81      \item \code{russelrao}
 82      \item \code{rodgerstanimoto}
 83      \item \code{cosine}
 84      \item \code{achiai}
 85      \item \code{carbo}
 86      \item \code{baroniurbanibuser}
 87      \item \code{kulczynski2}
 88      \item \code{robust}
 89
 90      \item \code{hamann}
 91      \item \code{yule}
 92      \item \code{pearson}
 93      \item \code{mcconnaughey}
 94      \item \code{stiles}
 95
 96      \item \code{simpson}
 97      \item \code{petke}
 98    }
 99    If the two fingerprints are of class \code{featvec} then the following methods 
100    may be specified: \code{tanimoto}, \code{robust} and \code{dice}.
101
102  }
103}
104\value{
105  Numeric value representing the distance in the specified metric between the
106  supplied fingerprint objects
107}
108\examples{
109# make a 2 fingerprint vectors
110fp1 <- new("fingerprint", nbit=6, bits=c(1,2,5,6))
111fp2 <- new("fingerprint", nbit=6, bits=c(1,2,5,6))
112
113# calculate the tanimoto coefficient
114distance(fp1,fp2) # should be 1
115
116# Invert the second fingerprint
117fp3 <- !fp2
118
119distance(fp1,fp3) # should be 0
120}
121
122\references{Fligner, M.A.; Verducci, J.S.; Blower, P.E.;
123  A Modification of the Jaccard-Tanimoto Similarity Index for
124  Diverse Selection of Chemical Compounds Using Binary Strings,
125  \emph{Technometrics}, 2002, \emph{44}(2), 110-119
126
127  Monve, V.; Introduction to Similarity Searching in
128  Chemistry, \emph{MATCH - Comm. Math. Comp. Chem.}, 2004, \emph{51}, 7-38
129}
130
131\keyword{logic}
132\author{Rajarshi Guha \email{rajarshi.guha@gmail.com}}