Help understanding wgsim_eval.pl output

Question

[Daniel Standage]

I used wgsim to simulate some Illumina reads, and then I mapped the reads back to the original sequence using several different aligners. The wgsim code distribution includes a wgsim_eval.pl script for, I presume, evaluating alignment accuracy. I'm getting output like this...

06x            0 / 149888              149888  0.000e+00
05x            0 / 29                  149917  0.000e+00
04x            0 / 83                  150000  0.000e+00
03x            0 / 0                   150000  0.000e+00
02x            0 / 0                   150000  0.000e+00
01x            0 / 0                   150000  0.000e+00
00x            0 / 0                   150000  0.000e+00

and this.

04x            0 / 138592              138592  0.000e+00
03x            0 / 529                 139121  0.000e+00
02x           18 / 6503                145624  1.236e-04
01x           11 / 1580                147204  1.970e-04
00x          207 / 1202                148406  1.590e-03

As far as I can tell the wgsim documentation doesn't describe the output format. Can anyone explain this output to me?

Answer 1

[Devon Ryan]

The output format is kind of strange. Anyway, the first column is just the MAPQ. The second is the number of correct alignments at that MAPQ. This is followed by a "/" and then the total number of alignments at that MAPQ. The fourth value is a cumulative sum of the total number of alignments. The final column is ratio of the cumulative correct alignments over the cumulative total alignments.

Perhaps you need to tweak the -g setting to get correct results...unless the aligner is really just not handling the reads well. You can also manually check where wgsim_eval.pl is expecting something to align by just looking at the read name. The format is chromosome_startForward_startReverse, where startReverse is the start position if the alignment should have bit 0x10 set.

[allrev]

Hi Devon, Thanks for your patient explanation. Sorry to say I was still a little bit confused by wgsim_eval.pl. I have the following question: 1) when should I and how can I use the option -p or -g 2) how can we explain their outcomes? 3) which column should I focus on if I want to get the mapping accuracy? 04x 0 / 138592 138592 0.000e+00 03x 0 / 529 139121 0.000e+00 02x 18 / 6503 145624 1.236e-04 01x 11 / 1580 147204 1.970e-04 00x 207 / 1202 148406 1.590e-03 (Is "the final column is ratio of the cumulative correct alignments over the cumulative total alignments." as you said)? 4) And how can I calculate the mapping accuracy? Looking forward to your reply. He