transcriptome pacbio alignment, Much N with Insertions

[luckylyw]

Hi,

I see the following somewhat weird data (SAM format) coming out of minimap2:

7964	16	NC_000913.3	835805	1	522S7=1293I4083N807=1D13=1X525=1I289=1D87=1X47=54S	*	0	0	TGTAC...
59048	0	NC_000913.3	835805	2	527S7=1269I4094N240=1X727=1D607=1X184=54S	*	0	0	TGTAC...

minimap2 was run with as
minimap2 -ax splice --eqx -uf -C5 NC_000913.3.fasta SRR628_candi_55.fasta > aln_55_candidate.sam
the input data is ecoli cDna, and the reference is NC_000913.3.fasta

In the alignment of transcriptome pacbio cDNA and reference, I found that when the N value is large, there will be a large number of consecutive insertion errors. What are the reasons for this phenomenon?

order  Qname              flag  map_quality  length  match  intro  mismatch  insert  delete 
25     49_75_2910_2826    2048  60           1398    1398   0      0         0       0
26     55_1_3509_3527     16    23           3527    1750   4112   0         1197    0 
26     55_1_3509_3527     16    23           3527    1750   4112   0         1197    0 

[tseemann]

Bacteria do not have introns so you should not use/need -x splice. Try -x map-pb instead ?

[lh3]

@tseemann is right that you should use -x map-pb for bacteria. Nonetheless, a cigar like 1293I4083N is still a problem with minimap2. I have seen this in other context. I need to fix this when I have time.