COMPARATIVE-ANALYSIS OF 7 MULTIPLE PROTEIN-SEQUENCE ALIGNMENT SERVERS- CLUES TO ENHANCE RELIABILITY OF PREDICTIONS

Motivation: The prediction reliability of seven multiple alignment ser vers currently available on the Internet (ClustalW, MAP, PIMA, Block M aker MSA, MEME and Match-Box) has been evaluated in terms of power (se nsitivity) and confidence (selectivity). Therefore, the alignments obt ained have been respectively compared to refined structural alignments for 20 families of related proteins with low levels of identity. Resu lts: Results clearly show that any powerful method remains reliable wh en the rate of identity falls. For some methods, power and confidence decrease linearly with the rate of identity, while other methods empha size reliability at the cost of a lower power. Increasing the number o f related sequences included in the alignment may either improve or de crease the quality of the predictions substantially. For some methods, the gain in power or in confidence is quite systematic;for others, th e effect of the addition of homologous sequences is highly unpredictab le. Extracting the consensus between two different methods may increas e the overall confidence of the predictions tremendously. Our conclusi ons induce users of sequence alignment methods on the Internet to sele ct the most suitable technique according to their requirements in term s of selectivity and sensitivity.