Variable results on variable minimal overlap

[krdav]

First of all, thanks for providing and alternative to PEAR which is faster and with a more permissive license.

Then to the problem: I have overlapping reads where I know the approximate overlap is 100, but this number can vary because of the variable length of antibodies. When I tried the merge function at first with the default settings of --overlap 15 I did not have very high success compared to PEAR. Increasing the minimum overlap to 100 actually increased the success by around 10% more reads merged. In the end I also have to lower the similarity to around 0.6 to get results consistent with PEAR.

So how come increasing the minimum overlap also increases the number of assembled reads? Do you have any experience that mitool merge is producing the same overlapping region as PEAR?

Thanks,
Kristian