Paired End Reads

[ahsen1402]

Hi I have some paired end read data, is there a specific treatment for them from the algorithm? How would you suggest I should deal with them?

Thanks

[mhalushka]

We do not have a specific treatment for paired end read data. I don't understand the utility of paired end read data for miRNAs and suggest you treat it all as non-paired data. Sorry.

[ahsen1402]

Thanks a lot. I used one end of the pairs to do the analysis but I got the following error"

Performing quantitation analysis of 317Uexo_CGGCTATG-ATAGAGGC_BH3WFNBCX2_L001_001.R1.fastq...
It takes: 630.38s

Performing annotation for all of the collasped sequences...
Alignment exited with none-zero status.

Any reason what might be the case?

Thanks

[mhalushka]

That error message means that no miRNAs were detected in the sample. miRge is not designed for regular RNA-seq as mature miRNAs will not be captured there. For miRNAs you need to start with small RNA-seq library preparation. I hope that helps.

[ahsen1402]

Thanks all our samples were small rna libraries. However, I had about 41 sample that I did analysis all together. When I switched and did the analysis one by one per sample instead of all 41 together then the analysis went through. Could it be also some memory related issue?

Eren

[mhalushka]

It may be a memory issue. We generally haven't done more than 20 samples at once if they sequencing depth is > 10 million reads each. I can't exclude that memory was a problem, but generally the "alignment exited with non-zero status" indicates that no miRNAs were found. Was it possible that the adapter removal step was set up correctly (not a typo) as that would also likely cause no miRNAs to be found. Either way, I'm glad you were able to get the samples analyzed.

[ahsen1402]

Hi Mark,

Just a suggesion after my experience today:

I got a new batch of samples and I was getting the same error even using a single sample, after increasing the memory size the job finished. It would be really good if there is a way the program can output memory error v.s. no miRNA detected. For that particular sample here is the resource summary:

  CPU time :                                   8008.00 sec.
    Max Memory :                                 24795 MB
    Average Memory :                             17752.14 MB
    Total Requested Memory :                     64000.00 MB
    Delta Memory :                               39205.00 MB

I hope this would be helpful to improve the software.

[mhalushka]

Thank you for the suggestion. We are looking into making that change. If we are able to do it, I will let you know.

[ahsen1402]

Thats great thanks a lot. A quick question in the annotation report what is the difference between all miRNA reads and filtered miRNA reads?

[luketerry]

@ahsen1402 Actually, in the step of annotation. mirge calls bowtie to align the reads to all kinds of RNA sequence libraries. Eventually, mirge tries to capture the status of bowtie running (a value is returned after calling bowtie). If the value equals to 0, it means bowtie runs successfully, otherwise, this error “Alignment exited with none-zero status” will show and mirge is forced to stop.
So the reason to cause this error message is that bowtie fails to run: 1) the bowtie index file can’t be found 2) the input fasta file can’t be found 3) there aren’t enough memories. I just tested made a test, if the input *.fasta file contained nothing, bowtie still can ran. So this means that even if there were no miRNA found, miRge should can work.
We have added a reminder in the part of how to use it in README.md file that "In order to make miRge2.0 run smoothly, setting the number of imput samples to be <= 20 and memory size to be >= 8 GB is recommended. Please keep in mind that predict module may take more memories" Thank you.