Abstract

This paper makes a quantitative comparison of different methods, called regularizers, for estimating the distribution of amino acids in a specific context, given a very small sample of amino acids from that distribution. The regularizers considered here are zero-offsets, pseudocounts, substitution matrices (with several variants), and Dirichlet mixture regularizers. Each regularizer is evaluated based on how well it estimates the distributions of the columns of a multiple alignment—specifically, the expected encoding cost per amino acid using the regularizer and all possible samples from each column. In general, pseudocounts give the lowest encoding costs for samples of size zero, substitution matrices give the lowest encoding costs for samples of size one, and Dirichlet mixtures give the lowest for larger samples. One of the substitution matrix variants, which added pseudocounts and scaled counts, does almost as well as the best known Dirichlet mixtures, but with a lower computation cost.

Citations

509	Henikoff J: Amino acid substitution matrices from protein blocks - Henikoff - 1992
436	Hidden Markov models in computational biology. Applications to protein modeling - Krogh, Brown, et al. - 1994
372	Detecting subtle sequence signals: a Gibbs sampling strategy for multiple alignment - Lawrence - 1993
253	A model of evolutionary change in proteins - Dayhoff, Schwartz, et al. - 1978
201	Profile analysis: detection of distantly related proteins - Gribskov, McLachlan, et al. - 1987
158	JM: The rapid generation of mutation data matrices from protein sequences - Jones, WR, et al. - 1992
128	Hidden Markov models of biological primary sequence information - Baldi, Chauvin, et al. - 1994
116	Amino acid substitution matrices from an information theoretic perspective - Altschul - 1991
80	Automated assembly of protein blocks for database searching - Henikoff, Henikoff - 1991
73	Position-based sequence weights - Henikoff, Henikoff - 1994
57	The classification of amino acid conservation - Taylor - 1986
56	D.: Using dirichlet mixture priors to derive hidden markov models for protein families - Brown, Hughey, et al. - 1993
56	Detection of conserved segments in proteins: iterative scanning of sequence databases with alignment blocks - Tatusov, Altschul, et al. - 1994
45	Automatic generation of primary sequence patterns from sets of related protein sequences - Smith, Smith - 1990
37	Improved sensitivity of profile searches through the use of sequence weights and gap excision - Thompson, Higgins, et al. - 1994
36	Weights for data related by a tree - Altschul, Carroll, et al. - 1989
33	Heniko , Amino acid substitution matrices from protein blocks - Heniko, G - 1992
19	A model of evolutionary change in protein - Dayho, Schwartz, et al. - 1978
19	Regularizers for estimating distributions of aminoacids from small samples - Karplus - 1995
15	Discovering sequence similarity by the algorithmic signi cance method - Milosavljevic - 1993
13	Some useful statistical properties of position-weight matrices. Computers and Chemistry 18(3):287{294 - Claverie - 1994
11	Heniko . Position-based sequence weights - Heniko, G - 1994
10	Automated assembly of protein blocks for database searching. NAR - Steven, Heniko - 1991
1	Weights for data related by a tree. JMB - Altschul, Carroll, et al. - 1989
1	Improved sensitivity of pro searches through the use of sequence weights and gap excision. CABIOS - Thompson, Higgins, et al. - 1994