# User:Marshall Hampton/ScoringMatrices

### From OpenWetWare

Line 8: | Line 8: | ||

where <math>p_i</math> and <math>q_i</math> are the background frequencies of the amino acids from two sets of proteins. In the vast majority of current treatments, the evolution of amino acid frequencies is considered symmetric in time and it is assumed that <math>p_i = q_i</math> for each amino acid i. In this article we will not make that assumption in order to develop scoring matrices for organisms in which the amino acid frequencies seem biased in time. | where <math>p_i</math> and <math>q_i</math> are the background frequencies of the amino acids from two sets of proteins. In the vast majority of current treatments, the evolution of amino acid frequencies is considered symmetric in time and it is assumed that <math>p_i = q_i</math> for each amino acid i. In this article we will not make that assumption in order to develop scoring matrices for organisms in which the amino acid frequencies seem biased in time. | ||

+ | |||

+ | As an example we will construct an asymmetric scoring matrix for the organisms <it>Plasmodium falciparum</it> and Saccharomyces cerevisiae. | ||

A good reference for the mathematics and statistics involved here is the article "Amino Acid Substitution Matrices from an Information Theoretic Perspective", J. Mol. Biol. 219, 555-565, 1991. | A good reference for the mathematics and statistics involved here is the article "Amino Acid Substitution Matrices from an Information Theoretic Perspective", J. Mol. Biol. 219, 555-565, 1991. |

## Revision as of 14:44, 9 September 2009

This will eventually be an article about constructing custom amino-acid scoring matrices using biopython. At the moment it is far from done.

## Introduction

A log-odds scoring matrix is constructed from some empirically found frequencies of single letter alignments *f*_{ij} from the formula

where *p*_{i} and *q*_{i} are the background frequencies of the amino acids from two sets of proteins. In the vast majority of current treatments, the evolution of amino acid frequencies is considered symmetric in time and it is assumed that *p*_{i} = *q*_{i} for each amino acid i. In this article we will not make that assumption in order to develop scoring matrices for organisms in which the amino acid frequencies seem biased in time.

As an example we will construct an asymmetric scoring matrix for the organisms <it>Plasmodium falciparum</it> and Saccharomyces cerevisiae.

A good reference for the mathematics and statistics involved here is the article "Amino Acid Substitution Matrices from an Information Theoretic Perspective", J. Mol. Biol. 219, 555-565, 1991.