Comparing the quality of de novo assembly from two runs' data

[NGS group]

We are using Miseq to sequence microbial genomes (~5Mb). In the first run (1#), library containing 12 samples were sequenced, and 6 samples were sequenced in the other run (2#). Obviously, the data of the 2# run have higher coverage depth that those of the 1# run. Therefore, I expected that the quality of de novo assembly data of the 2# run should be better than those of the 1#run. After analyzing by CLC Bio, I found that it is not true. Each sample from these two runs are shown as follow:

Sample of 1# run Sample of 2# run
N50 192439 261691
Max length 461804 450288
Average length 38501 8334
Count of contig 145 688

I dont understand why there are more contigs of 2# run that 1# run? Is that because of the high coverage? And how about the de novo assembly of these two runs?

Thank you very much!

[NGS group]

Is there anyone can help me????

[krobison]

Some of your question is specific to CLC bio, which I don't use.

What theoretical coverage did you actually achieve in the experiments? Are you using paired ends? What read lengths? What does the quality profile & insert size distribution look like for each? Are your samples very closely related, or perhaps not?

If the libraries are quite different & poorer libraries are concentrated in the second run, then that might explain things.

[NGS group]

Krobison, thanks for your reply. The coverage of the 2# run is around 140x. I used paired ends 2x250. The insertion size of the 2# run looks longer than that of the 1# run from Bioanalyzer data. The samples are very closely related.

Some of your question is specific to CLC bio, which I don't use.

What theoretical coverage did you actually achieve in the experiments? Are you using paired ends? What read lengths? What does the quality profile & insert size distribution look like for each? Are your samples very closely related, or perhaps not?

If the libraries are quite different & poorer libraries are concentrated in the second run, then that might explain things.