pvalue cutoff changes the predicted p-site, which is detected at site that is not the peak in metplot function

[dougbarrows]

Hello!

I am running a few BAM files through the metaplot function to obtain the p-site, and eventually identify ORFs. I am running version 1.2.15 of RiboCode. When I used a p-value cutoff to 0.01 (compared to the default of 0.001), a few more read lengths were included in the output for psite determination, which of course is expected, however the f0_sum, f1_sum, and f2_sum values changed. This also resulted in a different p-site for read lengths that were already identified with the default p value cutoff of 0.001, and the shifted psites are clearly no longer at the peak density in the histogram. The PDF histogram files look exactly the same for each pvalue cutoff as well. As I understand it, and after looking at the code here on github, it's surprising to me that the f0, f1, and f2 (and psite) values would change just by modifying the p-value cutoff. Is this expected behavior of this function?

Here are the text outputs for one of the BAM files for each pvalue cutoff:

p value cutoff of 0.001:
read_length proportion(per total mapped reads) predicted_psite f0_sum f1_sum f2_sum f0_percent pvalue1 pvalue2 pvalue_combined
30 13.14% 12 3650 284 545 81.49% 0.00014595245039721014 0.00014623915190606628 1.5078751056171674e-07

p value cutoff of 0.01:
read_length proportion(per total mapped reads) predicted_psite f0_sum f1_sum f2_sum f0_percent pvalue1 pvalue2 pvalue_combined
30 13.14% 15 3657 304 1235 70.38% 0.00014623915190606628 0.0020892861167162544 2.2541602661910716e-06
29 8.02% 12 1540 250 249 75.53% 0.00014509478281012493 0.001538816524417959 1.6164865319454036e-06