On-line tools for sequence retrieval and multivariate statistics in molecular biology

We have developed a World-Wide Web server for browsing sequencecollections structured under the ACNUC format and for performingmultivariate analyses on sequences. General collections (likeGenBank or EMBL), as well as specialized data banks (like Hovergenand NRSub) can be accessed. This system allows complex queriesto be constructed, and the result of each query, representedby a list of sequences, is stored on the server. It is thenpossible to reuse this list to compute multivariate analyseson the sequences. Two examples of applications are shown. Thefirst one consists in a study of codon usage with correspondenceanalysis on all the protein genes of Haemophilus influenzaeRd. This study allows the highly expressed genes and the integralmembrane proteins of this organism to be identified. The secondone consists in an ordering of 70 aligned protein sequencesof growth hormone with principal coordinate analysis. With thismethod, we are able to re-establish the patterns of relationshipsbetween the sequences previously determined with tree buildingprograms.