The PAM250 scoring matrix
ARNDCQEGH ILKMFPST WYV
A2
R-26
N002
D0-124
C-2-4-4-54
Q0112-54
E0-113-524
G1-301-3-105
H-1221-331-26
I-1-2-2-2-2-2-2-3-25
L-2-3-3-4-6-2-3-4-226
K-1310-510-20-2-35
M-10-2-3-5-1-2-3-22406
F-4-4-4-6-4-5-5-5-212-509
P10-1-1-30-1-10-2-3-1-2-56
S10100-101-1-1-30-2-313
T1-100-2-100-10-20-1-2013
W-62-4-7-8-5-7-7-3-5-2-3-40-6-2-517
Y-3-4-2-40-4-4-50-1-1-4-27-5-3-3010
V0-2-2-2-2-2-2-1-242-22-1-1-10-6-24

W. A Pearson, Rapid and Sensitive Sequence Comparison with FASTP and FASTA, in Methods in Enzymology, ed. R. Doolittle (ISBN 0-12-182084-X, Academic Press, San Diego) 183(1990)63-98.

The matrix is frequently used to score aligned peptide sequences to determine the similarity of those sequences. The numbers given above were derived from comparing aligned sequences of proteins with known homology and determining the "accepted point mutations" (PAM) observed. The frequencies of these mutations are in this table as a "log odds-matrix" where:

Mij = 10(log10Rij),
where Mij is the matrix element and Rij is the probability of that substitution as observed in the database, divided by the normalized frequency of occurence for amino acid i. All of the number are rounded to the nearest integer. The base-10 log is used so that the numbers can be added to determine the score of a compared set of sequences, rather than multiplied.