Minimap2: Gene fusion not detected

[twrzes]

Hi Heng,

First of all I would like to thank you for creating such a great software!

I have problem with minimap2 (I tested versions 2.7, 2.11 and r847) for detection of fusion genes. I generated some Nanopore cDNA sequencing on some human samples. I have a read which sequence can be found below:

>read1 TTGAAATTGCTTCGTTCGAATCAATTCCCAGATTGTTTTATGGTTCTTTACCTGTCACTCTATCTTCTTTTTTTTTTTTTTCCTTTTTTTTTTTTTTTGTTAATTAATTAATGCTTTTAGGACATAATAATAACAATTGAATGTCTGCACAACCCTTTGTTGACCATAAGCACCCCATGCAGCCAGTCACTGCACCTCCATCTACCACTCCTACCATGTTCCTCTACACAGCCGACTAGTTATGATCAGAGCAGTTACTCTCAGCCACCTATGGGCAACCGAGCAGCTATGGACAGCAGAGTAGCTATAGTCGCTCGAAGCAGCTATGGGCAGCAGCCTCCTAGTTACCACCCAAACTGGATCCTACAGCCAAGCTCCAAGTCAGCCAGCCAACAGAGCAGCAGCTACGGGCATGGGGAGTTCCTGCTGGCCTATATTAACCTCCACACCGACCAATCCCTCGCGATTGAGTGTCAAAGAAGACCCTTCTTATGACTCAGTCAGAAGAGACTTGGGGCAATAACACTTCTGGCCTCAACAAAGTCCTCCTTGGAGACACAAACGATCATTAGATACAGAGCACC

I am trying to align it against the reference genome (ENSEMBL v93). When I used BLAST or BLAT, I can see that this read overlaps EWSR1 (read positions 169-429, genomic location of the gene chr22:29,268,009-29,300,525) and FLI1 (read positions 419-582, genomic location of the gene chr11:128,686,535-128,813,267). However, when I aligned this read with minimap2, the output of SAM file is:

@PG ID:minimap2 PN:minimap2 VN:2.11-r797 CL:minimap2 -t 15 -p 0 -N 100 -ax splice hg38_genome.fa test.fa
read1 0 chr22 29282512 60 171S30M1D5M1I9M1D4365N1D49M4D47M1I27M3D7M1D3M1D33M1D30M171S * 0 0 TTGAAATTGCTTCGTTCGAATCAATTCCCAGATTGTTTTATGGTTCTTTACCTGTCACTCTATCTTCTTTTTTTTTTTTTTCCTTTTTTTTTTTTTTTGTTAATTAATTAATGCTTTTAGGACATAATAATAACAATTGAATGTCTGCACAACCCTTTGTTGACCATAAGCACCCCATGCAGCCAGTCACTGCACCTCCATCTACCACTCCTACCATGTTCCTCTACACAGCCGACTAGTTATGATCAGAGCAGTTACTCTCAGCCACCTATGGGCAACCGAGCAGCTATGGACAGCAGAGTAGCTATAGTCGCTCGAAGCAGCTATGGGCAGCAGCCTCCTAGTTACCACCCAAACTGGATCCTACAGCCAAGCTCCAAGTCAGCCAGCCAACAGAGCAGCAGCTACGGGCATGGGGAGTTCCTGCTGGCCTATATTAACCTCCACACCGACCAATCCCTCGCGATTGAGTGTCAAAGAAGACCCTTCTTATGACTCAGTCAGAAGAGACTTGGGGCAATAACACTTCTGGCCTCAACAAAGTCCTCCTTGGAGACACAAACGATCATTAGATACAGAGCACC * NM:i:23 ms:i:181 AS:i:145 nn:i:0 ts:A:+ tp:A:P cm:i:37 s1:i:168 s2:i:100 dv:f:0.0453
read1 272 chr1 166276810 0 171S26M1D34M1D4M1D8M3D26M1I49M4D49M2D9M1I4M1D28M174S * 0 0 * * NM:i:28 ms:i:165 AS:i:165 nn:i:0 tp:A:S cm:i:18 s1:i:100 dv:f:0.0694
read1 256 chr14 35881529 0 171S30M1D5M1I7M14I4M8I25M4D47M1I27M3D9M1D1M1D26M208S * 0 0 * * NM:i:48 ms:i:87 AS:i:87 nn:i:0 tp:A:S cm:i:11 s1:i:57 dv:f:0.0909

So as you can see the alignment to EWSR1 is there, bit the alignment to FLI1 is missing. Is there any parameter I can change (apart from -p 0 which I guess doesn't change much in terms of splitted alignments) or it isn't just possible to identify gene fusions in minimap2 alignments?

Thank you very much for all your help, I am very grateful for it.

Kind regards,
Tomasz Wrzesinski

[lh3]

Sorry for the late response.

The hits to FLI1 are too small and introns are quite long. Minimap2 has low sensitivity to such short hits. You can actually find hits to FLI1 with -x splice -k13, but probably it will be slow.

[lh3]

I am closing this for now. If you have followup questions, please reopen or post it to this thread. I will be notified.