Deletion at the end of cigar

[ebioman]

I encountered a strange results which I believe might be a bug in the cigar generation.

Aligning a cDNA sequence against 2 targets which is supposed to give a fusion transcript results as expected in a primary alignment with a supplementary one in the SA string.
The supplementary alignment has though at the end a deletion added which I cant explain:

NODE_1_length_906_cov_22.883314_g0_i0 16 CHS.55128.33 2810 60 427M479S * 0 0 TGGAAGGGGATGAGCACATTCACAGCGAAGTGAGTGAAAAAGTCCTGATGAATAGTGAGAGGCCTGATGAGAACCACAGTGAGCCCTCTCACCAGGACGTCATCAAGGTGAAGGAAGAATTTACAGACCCCACTTACGACATGTTTTACATGAGCCAGTATGGACTGTACAATGGTGGGGGTGCCAGCATGGCCGCCTTGCATGAGAGCTTTACATCGTCTCTGAATTATGGCAGCCCTCAAAAGTTCTCCCCAGAAGGTGACCTATGTTCTAGCCCAGACCCCAAAATCTGTTATGTGTGCAAGAAGAGTTTCAAAAGCTCCTACAGTGTGAAACTTCACTACAGGAACGTTCACTTGAAAGAGATGCACGTCTGCACAGTGGCTGGTTGCAATGCTGCATTCCCCTCTCGCCGAAGCCGAGACAGGTACCTGGTCCCTGAGCGCGGGCCTGAGCGAGGCCTGGGCAGCAAGTCCCTGCGCGCCTTCGTGGGGGAGGTGGGTGCCCGCTCTGCGGCCCCCGGGGGCGGCTCGGTGGCGGCGGCCGCTGCGGCCATGGGTGCGGCGCTGGGCTCCATGGTGGGCCTCATGACCTACGGGCGGCGCCAATTCCAGTCCCTGGACACGACGATGCGGCGCCTGATCCCGCCCTTCCGCGAGGCTTCGGCCAAGCTAACCACGCTGGTGGATGCCGACGCCGAGGCCTTCACCGCCTACCTGGCGCACGGCGGCCCTACAGGAGGGTCTGAGGCGGGCAGTCTCTGTGCCGCTGACGCTGGCGGAGACGGTGGCCTCGCTGTGGCCGGCCCTGCAGGAACTGGCCCGGTGTGGGAACCTGGCCTGCCGGTCAGACCTCCAGGTGGCGGCCAAAGCCCTGGAGATGGGCGTGTTTGGCGCATATTTCAAC * NM:i:0 ms:i:427 AS:i:427 nn:i:0 tp:A:P cm:i:80 s1:i:415 s2:i:0 de:f:0 SA:Z:CHS.35042.9,1311,-,427S479M44D,60,44; cs:Z::427 rl:i:0

So the cigar is 427S479M44D which is really bizarre with the 44 Deletions at the end.
The complementary alignment is

NODE_1_length_906_cov_22.883314_g0_i0 2064 CHS.35042.9 1311 60 427H291M44D188M * 0 0 GTACCTGGTCCCTGAGCGCGGGCCTGAGCGAGGCCTGGGCAGCAAGTCCCTGCGCGCCTTCGTGGGGGAGGTGGGTGCCCGCTCTGCGGCCCCCGGGGGCGGCTCGGTGGCGGCGGCCGCTGCGGCCATGGGTGCGGCGCTGGGCTCCATGGTGGGCCTCATGACCTACGGGCGGCGCCAATTCCAGTCCCTGGACACGACGATGCGGCGCCTGATCCCGCCCTTCCGCGAGGCTTCGGCCAAGCTAACCACGCTGGTGGATGCCGACGCCGAGGCCTTCACCGCCTACCTGGCGCACGGCGGCCCTACAGGAGGGTCTGAGGCGGGCAGTCTCTGTGCCGCTGACGCTGGCGGAGACGGTGGCCTCGCTGTGGCCGGCCCTGCAGGAACTGGCCCGGTGTGGGAACCTGGCCTGCCGGTCAGACCTCCAGGTGGCGGCCAAAGCCCTGGAGATGGGCGTGTTTGGCGCATATTTCAAC * NM:i:44 ms:i:308 AS:i:354 nn:i:0 tp:A:P cm:i:45 s1:i:413 s2:i:0 de:f:0.0021 SA:Z:CHS.55128.33,2810,-,427M479S,60,0; rl:i:0

Here, the deletion seems to be not towards the edge but internal. I guess it might be a bug in the reporting of supplementary alignments

I tried this with the original command I was using

 minimap2 -ac -k19 -w10 -A1 -B19 -O39,82 -E3,1 -s30 -z200 -N1 -t 5 --min-occ-floor=100 --cs ref.fasta query.fasta

but as well with the -ax asm5 flag which both generate that pattern
I verified that the 2 reference sequences have no N somewhere and the version is Minimap2-2.17 (r941)

[lh3]

SA doesn't give the right CIGAR. It only intends to show the coordinate.