Reads in Peaks Statistic is probably wrong

[alexg9010]

I have the feeling that there is something wrong with this.

[frenkiboy]

can you create a test data set where you know the percentage, and see how it performs?
I went through the code, and can't figure out the mistake

[alexg9010]

If I just define one peak, then the resulting plot shows what I would expect:

[alexg9010]

This is how it looks when two peaks are defined.

[messersc]

Has this been fixed? I would need to know for the discussion of the experiment with our collaborators.

[alexg9010]

Hi @messersc ,

Sorry, but this is not fixed yet. I can have a closer look at this next week, but I need to create a controlled test set first.

How does your distribution look like, do you have get some bars?

For the case your discussion is soon, please use this code to get some rough values:

library(dplyr)
library(ggplot2)

lstats <- readRDS("[/path/to/output]/Analysis/Summarized_Data_For_Report.RDS")

dd = lstats$Peak_Statistics$peaks_sample %>%
  dplyr::select(-bed_file, -bw_files, -bam_file, -sample_id,-library,-genome_type)                 %>%
  tidyr::gather(sample_cnt, value, -sample_name,-bam_name,  -mapped_total, -peak_number)  %>%
  mutate(value = as.numeric(value)) %>%
  mutate(mapped_total = as.numeric(mapped_total)) %>%
  mutate(value      = value/mapped_total)

g = dd %>%
  dplyr::filter(bam_name == sample_cnt)                          %>%
  ggplot(aes(bam_name, value, fill=sample_name)) +
  geom_bar(stat='identity', position='dodge',show.legend = FALSE) +
  xlab('Sample name')                           +
  ylab('Percentage of reads in peaks') +
  coord_flip() +
  theme(axis.text.x = element_text(angle = 45, hjust = 1)) + 
  scale_fill_discrete('Peak Name')

print(g)

Best,
Alex

[messersc]

Hi Alex,

wow, you're super responsive 👍

I just needed to know if we can rely on these numbers or not. I will try to run your code, maybe I can contribute a bit to find the bug.

Thanks for your help and hope you have a nice weekend.
Clemens

[alexg9010]

@messersc I figured out the bug that caused this issue. It had to do with some default settings in summarizeOverlaps that were messing with our counts.
I will soon draft a new release and then it should be available on guix quite fast.