Poor mapping in v1.3.20

[Jakob Nissen]

When upgrading to KMA 1.3.20 (see #38), the number of mapping reads suddenly decreases quite a bit.

In the attached plot, I map influenza reads to a closely related (92-99% nt identity) reference (composed of 8 individual genetic segments, labeled in the plot). The X-axis shows the relative position of each of the segments, the Y-axis the log10 depths. The first pane is KMA v 1.3.19, second pane in KMA v 1.3.20.

First note the weird dip in depth for PB1 with 1.3.19, this is what spurred me to open #38. But look at the results for 1.3.20! It’s worse nearly across the board.

I’m not 100% convinced that the former is more “correct”, but I’m surprised that 250 bp high-quality Illuimina reads maps so poorly to >90% identical sequences, so I assume this must be some kind of bug introduced in v 1.3.20.

[ptlcc]

Hi Jakob

You are definitely right, something has gone (completely) wrong in v1.3.20.
I will go through the changes again, a little more thorough this time.

Best,
Philip

[ptlcc]

Just to avoid too many immediate problems I have re-pushed v1.319 as v1.3.21.

[ptlcc]

Fixed Issue #40

→ <<cset ce82c6016c22>>

[ptlcc]

Hi Jakob

It should be fixed in the latest version (1.3.22). Can I get you to verify it with something like the diff.pdf.

Best,
Philip

[Jakob Nissen]

I can’t upload a file in a comment, it seems, so here it is as a bitmap-picture.

It looks like it worked! Not only for the noticable dip, but generally improved mapping across several segments.

Thank you, Philip!

‌