How To Select Best Blastp Hit For Kegg Annotation?

Question

[Naren]

BlastP Output:

Query    Hit     e-value    id  bit_score   score    qry_start    qry_end    hit_start    hit_end    align_len hit_len
protein    hit1    2.00E-06    49    55.1     131            11            73        2    63         63        169
protein    hit2    4.00E-05    22    50.8      120            12            188        9    239         234         361
protein    hit3    5.00E-05    37    50.4      119             9            82        7    81         75         493
protein    hit4    5.00E-05    37    50.4      119             9            82        7    81         75         493
protein    hit5    5.00E-05    37    50.4      119             9            82        7    81         75         493

By all other parameters except alignment length, hit1 is he best sequence hit of BlastP for my query protein.
But here, these hits are used for annotating kegg pathways.
I have only one doubt: If alignment length affects the selection of best hit then hit2 (alignment length= 234) might be a good hit. Is it so?
What parameters should be taken as selection criterion for best hit?

Thanks in advance.

Answer 1

[justinhaselbach]

Lot of subjectivism issues involved- though standard parameters could be adapted from Published literature.

However-

1. Bit Score, E-values, alignment lengths are usually "length/ match non-specific"! Imagine the following situations (1), very high E-value for 22 nts, or (2)for a 1000 bp ORF: an alignment length of 200 bp? Or a (3) high bit score over just 500 bp ?

2. Thus, most people will use a double-edged sword: "E-value < 1-E05 AND 80 % Coverage of length" (some may prefer 50 % as cut-off as well!) (=YOUR selection criterion for best hit) : And this stringency will affect your "gene call/ KEGG annotation" or all those numbers for sure !

Thanks,