Multiple mapping with longreads - Get primary mapping

Question

[ricardoguerreiro2121]

Hello,

I'm mapping scaffolded pacbio to a reference assembly with Minimap2. The scaffolds are very long, so I would expect mostly unique mappings. But that is far away from the case. I get the 4x more mappings than my original reads.

My questions are:

• Is this normal with such long scaffolds? Why does this happen, is it the Ns?
• Is this normal with unscaffolded pacbio assembly/ reads ? (also happens, see example in the end)
• How do I get only best/primary mappings?

I am aware of filtering bam files through the -q and -f parameters, though I don't know what I should give to -f.

Here, for example, is the mapping of my scaffold1, with a length of 105192

samtools view contigs_against_ref.bam  | grep "scaffold1,"  | cut -f1-5

 (scaffold name)   (FLAG)     (HIT)     (HIT_POS) (MAPQ)
scaffold1,105192    2048    NC_024468.2 10880006    60    
scaffold1,105192    2048    NC_024468.2 20011587    60    
scaffold1,105192    2048    NC_024468.2 35345645    29    
scaffold1,105192    2064    NC_024468.2 38040128    24    
(...)  
scaffold1,105192    2064    NC_024468.2 48840231    60          
scaffold1,105192    2048    NC_024468.2 58654417    19  
(....)
scaffold1,105192    2064    NC_024468.2 114658611   60
scaffold1,105192    2048    NC_024468.2 115053147   60
scaffold1,105192    2048    NC_024468.2 115053624   41
(...)
scaffold1,105192    2064    NC_024468.2 129707383   60
scaffold1,105192    256     NC_024468.2 143808174   0
scaffold1,105192    2064    NC_024468.2 144478215   22
scaffold1,105192    2048    NC_024468.2 144478299   60

Same thing using my raw pacbio reads, here is read "29" mapping many times, but according to the FLAG they are all supplementary alignments?

29  0   NC_024465.2 13819334    60     
29  2064    NC_024465.2 170243992   60   
29  2048    NC_024465.2 13741283    60
29  2048    NC_024462.2 88062987    60
29  2048    NC_024468.2 83284448    60
29  2064    NC_024463.2 12104985    1
29  2064    NC_024459.2 209527685   31

Appreciate your attention,
Ricardo

Answer 1

[WouterDeCoster]

It looks like your scaffolds are not mapping to multiple locations entirely, but rather have a non-linear alignment. The scaffolds contain structural variants (or errors) and as such get split in multiple alignments to the reference. So probably each part of a scaffold aligns once, but each scaffold gives rise to multiple alignments.

[ricardoguerreiro2121]

Thank you for clearing it up. Same when I use the raw pacbio reads? I expect Structural Variation, the reference is from a different genotype of my plant.

I want to see where my reads map in the reference, in order to clean out reads that don't belong to the target region that I am trying to assemble. So I'm not interested in the SVs just yet, though it is the long run objective to identify them!