View Single Post
Old 07-03-2014, 09:51 AM   #17
Brian Bushnell
Super Moderator
 
Location: Walnut Creek, CA

Join Date: Jan 2014
Posts: 2,707
Default

Imagine a read with this quality profile:

40, 40, 40, 40, 2, 2, 2, 2, 40, 2

What I would term "naive trimming" to Q10 would trim only the last base with quality 2, and stop because the next base has Q40. This would leave 4 internal bases with Q2, which is not desirable.

The Phred algorithm would trim the last 6 bases, because their average quality (calculated by summing the error probabilities) is 2.79, which is below 10. Trimming regions with average quality below a threshold gives the optimal result in terms of the ratio of retained bases to the expected number of errors.
Brian Bushnell is offline   Reply With Quote