Class TamuraNeiDistanceMatrix

  • All Implemented Interfaces:
    DistanceMatrix

    public class TamuraNeiDistanceMatrix
    extends BasicDistanceMatrix
    Date: 22/01/2006 Time: 17:28:59
    Version:
    $Id: TamuraNeiDistanceMatrix.java 996 2009-03-31 01:15:35Z amyzeta $ Estimation of the Number of Nucleotide Substitutions in the Control Region of Mitochondrial DNA in Humans and Chimpanzees. Koichiro Tamura and Masatoshi Nei, 1993 Estimated Distance is d = 2 (pi(A) pi(G) k_R + pi(T) pi(C) k_Y + PI(A)PI(C)) t, where k_R/k_Y are the rates of Purine/Pyrimidine respectivly. When distances grow large, the formulas break as estimates of number of transition/transversion becomes inconsistent, which results in negative logs. Returning "infinity" is not optimal as it breaks operations such as constructing consensus trees where distances for resampled sequences vary between (say) 2-3 and out "infinity" 10. As a workaround I try to reduce the number of transitions/transversions by the minimum amout which brings the estimates back to a consistent state.
    Author:
    Joseph Heled