| Autor: |
Henri Moereels, Paul J. Lewi |
| Rok vydání: |
1991 |
| Předmět: |
|
| Zdroj: |
TrAC Trends in Analytical Chemistry. 10:283-290 |
| ISSN: |
0165-9936 |
| Popis: |
The linear sequences of amino acids of proteins can be compared for similarity or dissimilarity by means of computer algorithms. Here we focus on dissimilarities between 34 protein sequences belonging to the class of G-protein coupled receptors which mediate in the transmission of signals in cells. Latent variables analysis represents the receptors in a space of reduced dimensionality, showing their overall geometrical structure. Cluster analysis gives a logical classification of the receptors, in particular, the order in which they may have evolved from common ancestors. A method of serial clustering in spaces spanned by progressively reduced numbers of latent variables is described. This method gives an estimate of the number of structural latent variables and of the minimal number of structural clusters that can be distinguished. From the matrix of dissimilarities between the 34 receptor sequences we identified 15 structural latent variables and 16 structural clusters. Of the latter, 15 can be related to pharmacologically distinct types and subtypes of receptors. The 15 structural latent variables account for the between-type variation in the data. The residual latent variables explain the within-type variation in the data. The latter is attributed to noise produced by the computer algorithm and arising from differences in the cell species from which the receptors have been isolated. |
| Databáze: |
OpenAIRE |
| Externí odkaz: |
|
