Few variants per segment bug.

[sicotte]

Line5696 There is a problem (divide by 0) when L_min and L_max are equal $cov_L_sum = (1 * ($cov_L_sum-$cov_L_min)/($cov_L_max-$cov_L_min)); $maf_L_sum = (1 * ($maf_L_sum-$maf_L_min)/($maf_L_max-$maf_L_min));

Which can happen when there is a single variant per segment. Solution if($cov_L_max>$cov_L_min) { $cov_L_sum = (1 * ($cov_L_sum-$cov_L_min)/($cov_L_max-$cov_L_min); } if($maf_L_max>$maf_L_min) { $maf_L_sum = (1 * ($maf_L_sum-$maf_L_min)/($maf_L_max-$maf_L_min+1e-10)); }

[carsonhh]

Not sure I want to fix it that way. If you have a zero here then your input data is insufficient, as it is not a per segment calculation, but rather all segments. I guess I would want it to die here with a more informative message. I don't want it to continue.

[sicotte]

A few lines above, the program skips segments with 0 variants.. next if(!$count);

you might want to switch this test to testing for at least one variant (also). ... and keep the fix I suggested (there were two typos.. here is the correct version)

if($cov_L_max>$cov_L_min) { $cov_L_sum = (1 * ($cov_L_sum-$cov_L_min)/($cov_L_max-$cov_L_min)); } if($maf_L_max>$maf_L_min) { $maf_L_sum = (1 * ($maf_L_sum-$maf_L_min)/($maf_L_max-$maf_L_min)); }

[carsonhh]

It should. The segments in question are not the user supplied segments, but rather discovery segments from large multimegabase windows. If they are all empty then there are problems with your data.