I am currently using MIREAP software to identify novel miRNAs from deep sequencing data. I am running into a problem and I would be grateful if someone could help me.
My experiment is looking at viral microRNAs in virally-infected vertebrate samples. I have filtered out reads mapping to the host genome so the information below relates to reads that mapped to the viral genome only. I have six different small RNA libraries representing two treatments (three libraries each): early infection and late infection. The early infection libraries are smaller (around 400,000 reads each) than the late infection libraries (around 3 million reads each) because there is less viral load at the early time point. I used MIREAP on all six libraries separately. It worked perfectly with the early libraries and I obtained about 25 novel miRNAs. But when I ran MIREAP on the libraries of the later time point (the larger libraries), I obtained zero miRNAs from all three libraries. This of course does not make sense as I would expect to obtain probably even more miRNAs from the samples with more viral transcripts. The MIREAP run was successful in that all three output files were generated but the .aln and .gff files were blank and the log file showed that the run was successful but zero miRNAs were predicted. I also took some of the miRNA sequences predicted from the early time point libraries and searched the late time point fasta files for these sequences and reads matching those sequences are there in high numbers so I cannot understand why zero miRNAs would be predicted from the larger libraries. I also tried MIREAP by concatenating all six libraries and mapping them to the genome and running these single files (representing all reads from all 6 libraries) through MIREAP and this also resulted in zero miRNAs. So it appears that in my case there is a significant penalty for having more reads such that MIREAP can no longer predict miRNAs (even though my “larger” libraries are not even really large because they are only the viral portion).
Does anyone have any insight into why this might be happening and what I can do to be able to predict miRNAs from my larger libraries?
I tried contacting the creator of MIREAP but my email was undeliverable so I think their email changed and a Google search did not identify a new email address for this individual.