Bowtie 2 and mismatch

usegalaxy.eu support
1

Hi all,
I’ve mapped nanopore reads against a reference sequence using bowtie 2. The reads mapped were recorded in a separate file. After that I check the read length from my target virus. I need to accept a quality of sequencing from Q8, so sequencing error can occur, introducing mismatch during the mapping. The longer the read, the more error will theorically be. How to be sure that longer reads of my target were not discarded at the mapping stage ? Is there any option to accept a number or a % of mismatches using bowtie2 (or other mapping tools) ?
Thank you in advance for your advices

2

Hi @Sebw

For any tool, most of the direct usage parameters are included in the Galaxy wrapper. That means if you have found a publication with similar data and a protocol you want to follow, the methods can be applied in Galaxy.

Full protocols are available here →

So … BWA-MEM are minimap2 are two choices in Galaxy supported with single-cell tutorials.

3

thank you for your reply, but my question was not for single-cell purpose.
My issue has evolve since even allowing mismatches it did not change the number of reads mapped.
But if I used BWA-MEM or minimap2, mapped read quantity increased.
Bowtie2 : 508 reads mapped
BWA-MEM : 4450
Minimap2 : 4143.
when I do a taxo assignment directly on fastq using Kraken2, 2189 reads are identified as my target. The other are completely different.
I’m just interested in identifying the size of the reads mapped to my reference sequence.
any idea ?

4

The tool is reporting the single best match.

This means

  • If a read already has a good match, sub-hits will not be reported, even if they qualify for the minimum mapping criterial.
  • If a read is a multi-mapper, or has large gaps, or a part of the read doesn’t map (contamination), or the two halves map to different regions (chimera), or similar it may not be reported ever by Bowtie2/BWA-MEM.
  • Also, a read can be not mapping to a genome, but still identified by Kraken, because the target references doesn’t have assembly resolution in that area yet.
  • Maybe other reasons but those are the usual cases.

If you are looking for weaker matches, I still think this isn’t the right tool choice. Others can have other opinions!

You could try using a tool like BLASTN for this. Then maybe examine the alignments for the reads not captured with the other tools. You could do that in a genome browser, or by parsing the alignment statistics. Maybe you’ll find a pattern.