Effect of changing the minimizer occurrence threshold interval

[egoltsman]

As a workaround to memory overruns you suggested in #38 to lower -U, which did allow the run to complete (I set it to 25,100). Separately, on a larger machine, I was also able to complete the default run, but that machine is not always available. So now I'm wondering how safe it is to use the modified -U values in the future. I compared the results, and there is a noticeable drop in the number of segments and the corresponding increase in their average length. Could you please explain why that is and what effect this parameter has when lowering the upper and the lower boundaries of the interval? If what I am seeing is loss of sensitivity (i.e. due to fewer minimizers picked), then I'd have to be cautious of that in the future. Thanks in advance!

>cat h1.h2.minimap._on_bigmem_i.GFA.stats
Number of segments: 248396
Number of links: 341923
Number of arcs: 683846
Max rank: 1
Total segment length: 2664944961
Average segment length: 10728.615
Sum of rank-0 segment lengths: 2339104873
Max degree: 2
Average degree: 1.377
------------------------------------------------------------------------
> cat h1.h2.minimap-U25,100.GFA.stats
Number of segments: 224545
Number of links: 309235
Number of arcs: 618470
Max rank: 1
Total segment length: 2617000485
Average segment length: 11654.682
Sum of rank-0 segment lengths: 2339104873
Max degree: 2
Average degree: 1.377