search

huangyh09 / brie

BRIE: Bayesian Regression for Isoform Estimate in Single Cells
https://brie.readthedocs.io
Apache License 2.0
41 stars 15 forks source link

Brie isoform expression quantification error #8

Closed s6juncheng closed 5 years ago

s6juncheng commented 7 years ago

Hi Yuanhua,

thanks for providing the tool for splicing analysis. I got an error when running brie for isoform expression.

brie -a AS_events/SE.gold.gtf -s sample.bam -f AS_events/human_features_v19.csv.gz -o AS_events/ -p 50
[Brie] loading annotation file... Done. [Brie] loading reads for 12504 genes with 50 cores... [Brie] [====================] 100.0% done in 58.5 sec. [Brie] running Brie for 25008 isoforms on 12504 genes with 50 cores... Traceback (most recent call last): File "/opt/modules/i12g/anaconda/3-4.1.1/envs/python27/bin/brie", line 11, in load_entry_point('brie==0.1.3', 'console_scripts', 'brie')() File "build/bdist.linux-x86_64/egg/brie/brie.py", line 268, in main File "build/bdist.linux-x86_64/egg/brie/models/model_brie.py", line 300, in brie_MH_Heuristic File "build/bdist.linux-x86_64/egg/brie/models/model_brie.py", line 159, in Iso_read_check IndexError: index 0 is out of bounds for axis 0 with size 0

It might because my bam file uses ensembl chromosome names (1, 2, 3...) while the annotation SE.gold.gtf is from gencode followed you instructions.

I tried to generate brie-factor with ensembl annotation, it gave warnings like "No PhastCons data for ENSG...in. Treated as Zero". As PhastCons file is from UCSC, I assume the problem is again because of chromosome names.

Besides, the same error of brie for isoform expression persist if I use SE.gold.gtf generated from ensembl and human_features generated from gencode (as I cann't generate human_features from ensembl, see above).

Is there a way to get around of this?

Best wishes, Jun

huangyh09 commented 7 years ago

Hi Jun,

Glad to hear from you and you are using BRIE.

Normally, BRIE can handle the the bam file that aligned to Ensembl reference, see here:

I assume that you aligned your fastq file to Ensembl genome. Could you point out which aligner and which reference file you used? So I could have a try with some example data.

I think you are right that brie-factor doesn't support the different chromosome id in annotation file and PhastCons file in bigWig format. See here.

I think the simplest thing you could do is to modify the chromosome id in the annotation file by simply removing chr, so that you could run brie with your bam file. During the meantime, I will try to upgrade the program to support the differences in chromosome ids.

Cheers, Yuanhua

s6juncheng commented 7 years ago

Hi Yuanhua,

the problem solved when I install with pip in python3. Before the installation was done with python2.7 from source that cloned from your master branch. Not sure whether it was because of python2, but glad that my problem solved. Thanks!

Cheers, Jun

huangyh09 commented 7 years ago

Hi Jun,

Glad that the issue has been solved. In my environment, both Python 2.7 and 3.3 works well, with BRIE v0.1.3. Earlier version may cause such problems.

Cheers, Yuanhua