Big differences between two alignment tools outputs: BWA vs NanoPipe

Question

[alexandru.bologa.marian]

Hello,

I recently aligned (mapped to reference) some nanopore reads from Drosophila melanogaster obtained with Oxford Nanopore MINion. My reads are filtered so that only those with a quality score higher than 10 remain. I aligned the fastq file against D. melanogaster reference genome v6.32 with BWA MEM and NanoPipe, both with default parameters. When I visualized the alignments outputs (bam files) with Tablet, I found that there are some big difference regarding the coverage of some chromosomes. For example, the chromosome 3L has a complete coverage when aligned with BWA MEM, but only half coverage when aligned with NanoPipe (I attached a picture with the comparison of this alignments). Consequently, the consensus I have obtained from the alignment using NanoPipe contains wide gaps in some regions. If I use the consensus sequence obtained from the alignment using BWA MEM, I obtain more complete sequences for Drosophila`s chromosomes.

In this case, my question is, what aligner should I use ? It is more appropiate to use NanoPipe with the default parameters instead of BWA MEM, given that my reads are generated by the MINion Nanopore ?

Thank you !

[GenoMax]

Ideally you should use minimap2 (LINK) and then compare with Nanopipe (if you have long reads that is).

[alexandru.bologa.marian]

Thank you ! I will try minimap2, I have downloaded it just a few days ago. I want to try bwa mem with -ont2d option too, to see if there is a difference.

[ATpoint]

bwa mem was not developed with the long and especially error-prone reads from Nanopore in mind, therefore it is not unexpected that it does not perform super well. minimap2 is a more recent and recommended alternative from the same developer.

Answer 1

[alexandru.bologa.marian]

Hi !

I tried to work with minimap2, both on the command line and through the CLC platform. Differences between mapping to reference continue to exist, such as the number of reads associated with each chromosome and the chromosomes regions covered.I attached one more picture with a comparison between NanoPipe and Minimap2 (via command-line and CLC platform), using same reads and reference genome (D. melanogaster - chromosome 3L in picture). Which of the alignments is more reliable and why?

Thanks !

Picture: NanoPipe_vs_Minimap2