How accurate are PAM distances?

Joe Felsenstein joe at GENETICS.WASHINGTON.EDU
Mon Aug 30 16:24:26 EST 1993


Paul Berti raises the question of how well PAM distances between proteins will
behave for long evolutionary divergences.

Not very well, I would assume.  The problem is not the details of correction
for multiple hits.  It is that evolutionary rates will typically vary a great
deal from position to position, and this variation is not taken into account
in the PAM distance calculation.  If evolutionary rates vary among positions,
there will be more superimposed replacements than the PAM distance calculation
assumes.  Thus the distances will tend to miss these replacements, and thus
be undercorrected, to be smaller than they should be.

A correction such as that done by Jin and Nei (MBE 1990) would help, but it
is not currently available in my programs.  I hope to add it in the next
version.  Even then it requires the user to specify a parameter for the
degree of variation of evolutionary rate, and Berti would have to assign the
proper value to that parameter to get the right correction.

-----
Joe Felsenstein, Dept. of Genetics, Univ. of Washington, Seattle, WA 98195
 Internet:         joe at genetics.washington.edu     (IP No. 128.95.12.41)
 Bitnet/EARN:      felsenst at uwavm



More information about the Mol-evol mailing list