Assessment of similarities of pairs and groups of proteins using transformed amino-acid-residue data

Abstract

Using as a primary standard a representative set of 208 proteins whose amino-acid-residue mole frequencies have been accurately established, a set of standard distributions of mole frequencies is defined for each amino acids, in terms of which percentile values for the observed mole frequencies of the amino-acid residues in any other protein can be determined. Data so transformed have a distribution much closer to Gaussian than untransformed values, and allow meaningful determinations of correlations between the amino-acid-residue compositions of two proteins as well as between pairs of amino-acid-residues within groups of proteins. Of the 153 possible pairs of amino acids (Asx and Glx are used) 39 are significantly correlated at p less than or equal to 0.01 and 22 at p less than or equal to 0.001. A percentile table is included for those wishing to use the method with programmable calculators. The transformed data for amino-acid compositions have been used to perform principal components analyses on groups of proteins in order to determine if meaningful sub-groupings (observable clusters in scatter diagrams) were detectable. Such analyses are shown for the representative set of proteins and for a group of 184 globins. With regard to the globin chains, a correlation is observed for alpha chains in the first principal component projection (PCP), (accounting for 22% of the variance) with respect to the evolutionary time-scale while beta chains show such a correlation in the first and second PCPs (22% and 18% of the variance respectively). Thus, alpha and beta chains appear to diverge from a common progenitor, similar in position to globin chains from "primitive" forms. Furthermore, globins from "primitive" forms are nearer to one another than they are to globins from the vertebrates, a finding without a priori reason, suggesting perhaps that once a chain has reached a stable relationship with its environment, strong constrains are placed on the co-existing globin chains so that they maintain appropriate interaction with one another. In addition, positions of the epsilon, gamma and delta chains are in the order: epsilon (embryonal) more primitive than gamma (foetal) more primitive than delta equal to beta (adult).

Spontaneous deletion mutants of the Lactococcus lactis temperate bacteriophage BK5-T and localization of the BK5-T attP site

“…FASTA analysis also indicated similarity between ORF1904 and an unidentified ORF (ORF35) from the lactococcal bacteriophage bIL67 (19) and less similarity to the lactococcal lytic bacteriophage US3 lytic enzyme (16) and the Bacillus subtilis xylose isomerase (22). Analysis by COMPARE (18) indicated similarity between ORF1904 and numerous proteins involved in binding and/or degradation of cell wall glycoproteins. These sequence similarities and the observation that a number of cell wall-lytic en-zymes contain repeated sequence motifs (5,8,10,12) suggest that ORF1904 may be involved in cell lysis during the lytic cycle of BK5-T or in cell wall hydrolysis to enable phage DNA injection.…”

Identification of prophage genes expressed in lysogens of the Lactococcus lactis bacteriophage BK5-T

“…5). ORF63, however, showed no significant homology to Cro or any other characterized proteins, as determined by COMPARE (24) or FASTA (18) comparison with proteins in the GenBank database. The amino acid sequence of ORF63 was analyzed for the presence of possible helix-turn-helix structures, but no sequence fulfilling all the requirements of a helix-turn-helix motif was identified.…”

The MTX package of computer programmes for the comparison of sequences of nucleotides and amino acid residues

“…The method for deriving the correlation statistic (r) for amino acid residues is based on a concept developed previously [5] which utilizes the amino acid residue composition of a standard set of proteins [6] and which uses a table look-up to perform a percentile transformation. As a result, the analysis is based on a near normal distribution and a meaningful estimate of the probability for a given "r" can be given.…”