Entering edit mode
2.9 years ago
vinishavvenugopal
▴
30
Hi,
I'm getting the below error when I try to generate genome indices.Should the input files(reference file) be present in the same location as STAR? I have given absolute path for the reference file and the drive has more than 100gb of disk space, it still throws me an error. I could see that it started generating indices, but it did not complete the entire run. Any help would be appreciated.
Also why do I get the warning,"could not move Log.out", I have read write permission to that folder.
Thank you.
--Vinisha
@Vinisha please copy and paste the actual error in future. It is difficult to decipher screenshots.
It appears that you are either using an index what was created with a mismatched STAR version (
STARindexes have changed overSTARversion releases) or the index is corrupt in some way. You should re-createSTARindex and try again.You also appear to be using a directory that is mounted via
globus? Is that correct? Not sure if you can use those for high speed read/writes.Thanks Max.
Sorry about the screenshot, I will copy paste it next time.
You are right, it's not the current reference genome file. It's hg37.75 reference file, in order to generate indices for this should I use a older version of STAR?
And yes, the directory is mounted. and that's the only disk space I have more than 100gb(I have limited resource).
@vinisha it is fine to use an older genome build with
STARto create reference indexes. I thought you were using pre-created indexes but that is not the case so please ignore that comment.I think your problem is that
globusmounted directory. I suggest that you try switching to a local directory for genome index generation.STARalso needs 30-40G of free RAM for human genome so make sure you are allocating enough RAM.GenoMax Yeah I can try free up space in the local directory for generating indices, but for mapping, can I use the mapped drive
Globus? Is there any problem in using the mapped drive to get input from and also to store the output?My input files are about 200gb and I can't have them in local drive(doesn't support large volume of data), that's the reason I have a mapped drive.
You are going to need to try and see. Hopefully read mapping will work using the mapped drive. You will find out reasonably quickly. Is there no local scratch space available for use while the jobs are running? Are you using something like
cyverse(shared compute)?Yeah there's some space available in the local disk. I was able to generate indices having the data in local disk(now I'm left with less than a Gb free space). Do I need more space for mapping as well?
And I don't know what a cyverse is (I'm pretty new to Bioinformatics). The mapped drive is just a shared network drive that holds 30Tb of data.
You will need plenty more space for the mapping. Move the index to shared disk and see if you can write the alignments out to local disk. May need to do one sample at a time. This assuming the mounted drive does not work for alignments.