xpore icon indicating copy to clipboard operation
xpore copied to clipboard

diffmod.table missing majority of positions

Open JannesSP opened this issue 1 year ago • 5 comments

Dear xpore developers,

I ran xpore on two SARS-CoV-2 samples to compare variant differences. After running xpore dataprep and then xpore diffmod I only receive information for about 2500 out of 30000 genomic positions. I followed your guide to prepare the bam and nanopolish files. Then I used only default parameters for the dataprep like this: xpore dataprep --eventalign ventalign_result.csv --out_dir /dataprep --n_processes 12 Do you have any idea what might went wrong? Or it this a expected behavior for the uneven and special coverage that Coronaviruses have?

Kind Regards, Jannes Spangenberg

JannesSP avatar Apr 18 '23 12:04 JannesSP

Hi @JannesSP,

The reason why you are getting 2,500 out of 30,000 genomic positions is that xpore filters out the sites that do not have a significant difference between the two conditions.

Another thing that could have happened is that you included the following in the config.yml file to xpore diffmod:

method:
    prefiltering:
        method: t-test
        threshold: <P_VALUE_THRESHOLD>

This does prefiltering, which significantly lowers the number of sites included in xpore diffmod output.

Thanks!

Best wishes, Yuk Kei

yuukiiwa avatar Apr 19 '23 07:04 yuukiiwa

Hi @yuukiiwa,

I did not include a prefiltering in the config.yml. So it seems that xpore filters a lot of positions. Is it possible to disable this filtering?

Kind Regargs, Jannes Spangenberg

JannesSP avatar Apr 20 '23 12:04 JannesSP

Hi @JannesSP,

Sorry for the delayed reply!

I will talk to @ploy-np to potentially push a release with a flag to skip the filtering.

Thanks!

Best wishes, Yuk Kei

yuukiiwa avatar Apr 26 '23 00:04 yuukiiwa

Hello, I met the same problem and I modified the code to check the number of the position that will be used to compare in diffmod.py . I found that most sites will not join the comparison, which is not related to the filter. Can you tell me that how to solve this problem

JeremyQuo avatar Oct 02 '23 07:10 JeremyQuo

Hi @JeremyQuo,

xpore diffmod does internal filtering of sites if the distributions of the two conditions are overlapping too much from each other. @ploy-np and I are still considering whether to skip the filtering as it would produce too much noise in the diffmod.table.

Thanks!

Best wishes, Yuk Kei

yuukiiwa avatar Oct 12 '23 06:10 yuukiiwa