how to verify large number corrected reads is correct?

[sam789123]

Hi all,

I wrote a program of correction to correct pair-end reads.
But the method of verifying corrected reads is difficult for me.
I could only observe few corrected reads in the same time though blasting them and check the Identities.
I was wondering if there was a way to observe large number corrected reads.

Thanks.

[Brian Bushnell]

BBMap can produce output that is helpful for this kind of thing:

Pairing data:           pct reads       num reads       pct bases          num bases

mated pairs:            100.0000%            1000       100.0000%             300000
bad pairs:                0.0000%               0         0.0000%                  0
insert size avg:          270.70


Read 1 data:            pct reads       num reads       pct bases          num bases

mapped:                 100.0000%            1000       100.0000%             150000
unambiguous:             98.4000%             984        98.4000%             147600
ambiguous:                1.6000%              16         1.6000%               2400
low-Q discards:           0.0000%               0         0.0000%                  0

perfect best site:       24.4000%             244        24.4000%              36600
semiperfect site:        24.4000%             244        24.4000%              36600
rescued:                  0.0000%               0

Match Rate:                   NA               NA        98.2567%             147385
Error Rate:              73.3000%             733         1.6680%               2502
Sub Rate:                73.3000%             733         1.6680%               2502
Del Rate:                 0.0000%               0         0.0000%                  0
Ins Rate:                 0.0000%               0         0.0000%                  0
N Rate:                  11.3000%             113         0.0753%                113


Read 2 data:            pct reads       num reads       pct bases          num bases

mapped:                 100.0000%            1000       100.0000%             150000
unambiguous:             98.7000%             987        98.7000%             148050
ambiguous:                1.3000%              13         1.3000%               1950
low-Q discards:           0.0000%               0         0.0000%                  0

perfect best site:       22.0000%             220        22.0000%              33000
semiperfect site:        22.0000%             220        22.0000%              33000
rescued:                  0.0000%               0

Match Rate:                   NA               NA        98.2627%             147394
Error Rate:              75.0000%             750         1.6660%               2499
Sub Rate:                75.0000%             750         1.6660%               2499
Del Rate:                 0.0000%               0         0.0000%                  0
Ins Rate:                 0.0000%               0         0.0000%                  0
N Rate:                  10.7000%             107         0.0713%                107

That gives you the exact number reads and bases with errors.

It also has a few other useful flags - such as ehist and mhist - that produce histograms showing error rate distribution.