GoekeLab/xpore

Empty diffmod.table file, only header generated when running xpore diffmod after preprocessing my runs

nickst85 opened this issue · 5 comments

Hi,
I followed your guide and I successfully preprocessed my eventalign files of my two conditions.
I have one repeat per condition.
Then I run the command
xpore diffmod --config $CONFIGFILE/my_config.yml --n_processes 32

the config file was:

data:
Patient:
rep1: /scratch/xPore/data/patient-rep1/dataprep
CONTROL:
rep1: /scratch/xPore/data/CONTROL-rep1/dataprep

out: ./out_condition1_FdR

method:
prefiltering:
method: t-test
threshold: 0.1

The diffmod.log starts with
--- DIFFMOD ---
has 91800 lines like the following
ENST00000000442.11|ENSG00000173153.17|OTTHUMG00000150641.9|OTTHUMT00000319303.3|ESRRA-201|ESRRA|2274|protein_coding|
--- SUCCESSFULLY FINISHED ---

Can you understand what's going on?

Thanks,

N

Hi @nickst85,

The first column of your eventalign.txt file has to be just the ENSTxxxxx. You will have to modify the eventalign.txt from currently:

ENST00000000442.11|ENSG00000173153.17|OTTHUMG00000150641.9|OTTHUMT00000319303.3|ESRRA-201|ESRRA|2274|protein_coding|

to the following:

ENST00000000442.11

Do you mind also providing the fasta and gtf files used for the processing, please?

Thanks!

Best wishes,
Yuk Kei

Hi Nick (tagging you here @nickst85),

You can use the following script (not tested) to change the first column of the eventalign.txt file (please change the paths to make sure you have enough storage space for the new modified_eventalign.txt file):

outfile=open('modified_eventalign.txt','w')
file=open('eventalign.txt','r')
outfile.write(file.readline())
for ln in file:
    ln=ln.split('\t')
    outln=[ln.split('|')[0]]+ln[1:]
    outfile.write('\t'.join(outln))
outfile.close()
file.close()

If you prefer to run nanopolish eventalign instead, we would suggest you using the ENSEMBL reference fasta and gtf files (we are trying to fix the problem with gencode reference on our end).

Thanks!

Best wishes,
Yuk Kei

Hi @nickst85,

I maintain that too. Will follow on this issue: GoekeLab/m6anet#147

Thanks!

Best wishes,
Yuk Kei