miRDeep2: mapping produced bowtie.log and mapper.log - which one to trust?

Question

[CandiceChuDVM]

I am using miRDeep2 to analyze small RNA-Seq data.
When running mapper.pl, two log files were produced in the system: bowtie.log and mapper.log.

Before executing the bowtie function to map reads to genome index, mapper.pl has done (1) parsing fastq to fasta format (2) discarding sequences with non-canonical letters (3) discarding short reads (4) collapsing reads.

The bowtie.log looks like below:

# reads processed: 138945
# reads with at least one reported alignment: 74938 (53.93%)
# reads that failed to align: 50691 (36.48%)
# reads with alignments suppressed due to -m: 13316 (9.58%)
Reported 122526 alignments to 1 output stream(s)

However, the Mapping statistics of mapper.log shows:

 parsing fastq to fasta format
 discarding sequences with non-canonical letters
 discarding short reads
 collapsing reads
 mapping reads to genome index
 trimming unmapped nts in the 3' ends
 Log file for this run is in mapper_logs and called mapper.log_33416
 Mapping statistics

 #desc  total   mapped  unmapped    %mapped %unmapped
 total: 11965391    11469171    496220  0.959   0.041
 seq: 11965391  11469171    496220  0.959   0.041

In the Mapping statistics, the total column equals to the number of fastq reads in the original fastq file.

I was wondering which mapping statistics should I trust and what's the difference between these two?

[Nicolas]

Hi, Im having the same dilema, did you came to a conclution that you can share? or maybe another tool you recomend for aligning miRNAseq reads

best regards.

Answer 1

[afb]

Hi, an answer to this question was provided by Devon Ryan in this post: Understanding miRdeep2 mapper output. In case you did not come across this yet. Best of luck!