BLAST

This is the Title of the Book, eMatter Edition

Copyright © 2012 O’Reilly & Associates, Inc. All rights reserved.

106

|

Chapter 7: A BLAST Statistics Tutorial

Recall that calculating a sum score also requires you to calculate the effective length,

n´, of the subject sequence. To do so, use the Perl function,

effectiveLengthSeq,

given earlier in the chapter, because it also applies to the subject sequence for pur-

poses of calculating a sum score.

Calculating a Sum Score

If you look at Example 7-2, you’ll see that these two BLASTX HSPs comprise an

ordered set. In other words, these two alignments suggest that the query sequence

contains a minus strand gene, at least two exons of which are homologous to the

subject sequence. Because these two alignments comprise a consistently ordered set,

you will calculate their pair-wise ordered sum score. Using the Perl function

sumScore

that’s in the earlier section “The Sum Score,” the sum score for these two HSPs is

about 53 nats, or 77 bits.

Calculating the Pair-Wise Sum P-Value

The sum score (77 bits) for these two HSPs isn’t much more than that of the first

HSP’s individual bit score (71.2). Why, then, is the resulting Expect (2) for these two

HSPs so low: Expect (2) = 1e

-15

, when the first HSP with only a slightly lower bit

score has a much less significant individual Expect of 3.7e

-10

?

The reason for the discrepancy is that the familiar Karlin-Altschul equation ...