Question

In the original implementation of PSI-BLAST, the algorithm performed a multiple sequence alignment and deleted all but one copy of aligned sequence segments having ≥ 98% identity. In a recent modification, the program now purges segments having ≥ 94% identity. What do you think would happen if this percentage were adjusted to ≥ 75% identity. How could you test this idea in practice?

Answer 1

ANS) PSI-BLAST is an iterative application to go looking a database for proteins with distant similarity to a question sequence. we investigated over a dozen adjustments to the methods utilized in psi-blast, with the aim of enhancing accuracy in locating real positive fits. to assess overall performance we used a set of 103 queries for which the proper positives in yeast were annotated by using human professionals, and a popular measure of retrieval accuracy (roc) that can be normalized to tackle values among zero (worst) and 1 (quality). the adjustments we keep in mind novel enhance the roc score from 0.758 ± zero.1/2 to zero.895 ± 0.003. this doesn't encompass the blessings from four modifications we included in the ‘baseline’ model, even though they were not carried out in psi-blast version 2.zero. the development in accuracy became confirmed on a small 2d check set. this take a look at involved analyzing 3 protein households with curated lists of actual positives from the non-redundant protein database. the modification that accounts for the general public of the development is the use, for every database collection, of a position-precise scoring gadget tuned to that series’s amino acid composition. the usage of composition-based totally information is specially beneficial for massive-scale automated packages of psi-blast.