unexpected cigar string error

[Overcraft90]

Hi there, I'm working with three human genomes which I'm aligning to the CHM13 reference for then plotting the results with SyRi (v.1.6.3). For some reason, I'm persistently getting the following for both hap1 and hap2 of those three samples

Reading BAM/SAM file - ERROR - Incorrect CIGAR string found. CIGAR string can only have I/D/H/S/X/=. CIGAR STRING: 244022067S76871671N
Reading BAM/SAM file - ERROR - Incorrect CIGAR string found. CIGAR string can only have I/D/H/S/X/=. CIGAR STRING: 245130294S49682284N
Reading BAM/SAM file - ERROR - Incorrect CIGAR string found. CIGAR string can only have I/D/H/S/X/=. CIGAR STRING: 250382078S60062191N

This appears to be related to #180; however, as opposed to all issues concerning the use of --eqx when working with minimap2 I'm already doing so. Hence, I cannot pinpoint the cause of this issue... any help is greatly appreciated, thanks in advance!

for h in "${HAP[@]}"; do
	minimap2 -ax asm20 -t 192 --eqx path/to/CHM13_no-Y.fna.gz path/to/HG002v1.1.${h}.fasta.gz \
 	| samtools sort -@ 192 -O BAM - > path/to/${h}/CHM13_HG002-${h}.bam
	samtools index -@ 192 path/to/${h}/CHM13_HG002-${h}.bam
	minimap2 -ax asm20 -t 192 --eqx path/to/HG002v1.1.${h}.fasta.gz path/to/H9v0.1.${h}.fasta.gz \
 	| samtools sort -@ 192 -O BAM - > path/to/${h}/HG002_H9-${h}.bam
	samtools index -@ 192 path/to/${h}/HG002_H9-${h}.bam	
	minimap2 -ax asm20 -t 192 --eqx path/to/H9v0.1.${h}.fasta.gz path/to/RPE1v1.1.${h}.fasta.gz \
 	| samtools sort -@ 192 -O BAM - > path/to/${h}/H9_RPE1-${h}.bam
	samtools index -@ 192 path/to/${h}/H9_RPE1-${h}.bam
	
	source /hpc/apps/miniconda3/24.4.0/bin/activate syri_env
		syri -c path/to/${h}/CHM13_HG002-${h}.bam -r path/to/CHM13_no-Y.fna.gz -q path/to/HG002v1.1.${h}.fasta.gz -F B --dir path/to/${h} --prefix CHM13_HG002-${h}_ --nc 192  &
		syri -c path/to/${h}/HG002_H9-${h}.bam -r path/to/HG002v1.1.${h}.fasta.gz -q path/to/H9v0.1.${h}.fasta.gz -F B --dir path/to/${h} --prefix HG002_H9-${h}_ --nc 192  &
		syri -c path/to/${h}/H9_RPE1-${h}.bam -r path/to/H9v0.1.${h}.fasta.gz -q path/to/RPE1v1.1.${h}.fasta.gz -F B --dir path/to/${h} --prefix H9_RPE1-${h}_ --nc 192

		plotsr \
			--sr path/to/${h}/CHM13_HG002-${h}_syri.out \
			--sr path/to/${h}/HG002_H9-${h}_syri.out \
			--sr path/to/${h}/H9_RPE1-${h}_syri.out \
    			--genomes path/to/genomes_${h}.txt \
    			-S 0.75 \
    			-d 600 \
    			-o path/to/${h}/output_plot-v2.pdf
	conda deactivate
done

[lrauschning]

Hi @Overcraft90,
judging by your cigar string, it looks like minimap2 is not finding any alignment between your genomes.
I'd advise

1. checking your input files, and parameter expansion in the shell script
2. checking minimap2's parameters (do you need 192 cores? is asm20 appropriate for your genomes?)
3. running a single Chr or another small test case for troubleshooting
   Also, syri can work directly with minimap2's PAF files, there's no need to convert it to BAM unless you are also using other software on the alignment.
   Cheers,
   Leon

[Overcraft90]

@lrauschning absolutely, I agree with you for point 2.

Concerning the asm20 considering we are talking about human genomes!!! However, initially that value set at asm5 was giving another error about not finding alignment for chr, which changed with different runs... so, that's why I thought about increasing it.

Still, I'm not expert but how more threads could hinder the CIGAR string formation? I will drop to 16/24 if this solves the problem; in another (successful) experiments I used SyRi on CHM13 and H9 with 96 threads and asm5 and everything went well, so I'm also confused.

If the problem persists after a few more experiments I will switch to native PAF from minimap2. Thanks for the quick answer!

[Overcraft90]

@lrauschning I have an update. I tried running SyRi on PAF files and got the following:

IndexError: list index out of range

which in several issues has been reported as chromosomes in different assemblies being oriented with opposite strand. So, as suggested in #254 I tried fixchr but it prompted this

2025-09-22 14:44:05,278 - fixchr - INFO - Reading alignments (fixchr.py:68)
Reading Coords - ERROR - Error in reading the alignment filelist index out of range

after running the following command:

./bin/fixchr -c /path/to/T2T_HG002_hap/hap1/CHM13_HG002-hap1.paf -r /path/to/CHM13_ref/syri/CHM13_no-Y.fna.gz -q /path/to/T2T_HG002_hap/HG002v1.1.hap1.fasta.gz -F P

How can I fix chromosomes' orientation (hoping this is the actual cause of the issue)?

[lrauschning]

Indeed, differing chromosome orientation could plausibly cause minimap2 not to find any alignment (as indicated by the indexerror).
You can try making dotplots of each chr, or running a few chromosomes individually to see which are in the wrong orientation.

Looking at the CIGAR string in the OG issue again, it seems that the query sequence is a lot shorter than the reference in all error messages (244Mb vs 77 Mb, 245 Mb vs 50 Mb, 250 Mb vs 60 Mb).
Are you sure you are aligning the same chromosomes, your assemblies are chromosome-scale/correctly scaffolded and the haplotypes are balanced ?

[Overcraft90]

@lrauschning got it! Yes, I kind of get to the same conclusion; definitely I'm aligning the same chromosomes since these are haplotype-resolved (T2T or near T2T) assemblies.

For the above they are chromosome-scale and correctly scaffolded (all assembled with Verkko); however, I'm aware only two of them are produced in-house (aside from the reference in CHM13), while the problematic one – HG002v1.1 from here – has been generated by another team... so, I cannot guarantee the balance between haplotypes across individuals.

[mnshgl0110]

The error from syri seems to imply that minimap2 didn't find any alignments in the sequences.

You can visualise the alignments using https://dgenies.toulouse.inra.fr/

For human genomes, I would expect a clear diagonal.

Also, the chromosome sizes (based on cigar) are very suspicious. You can check the chromosome size using samtools faidx. Homologous chromosomes should have similar sizes. I don't think such a large size difference in chromosomes could exist in nature. So, I would guess that either the assemblies consists of scaffolds, or something else is wrong with them.

[Overcraft90]

@mnshgl0110 thanks I will check them out, I would expect also a diagonal without much variation in it. Very fast and intuitive way to do it, I didn't think about it, but this is the result of samtools faidx for the genomes used in the analysis:

chr	H9_1	H9_2	RPE1_1	RPE1_2	HG002_1	HG002_2
chr1	250382078	243198671	245130294	246029683	244022067	244022067
chr2	242438398	242142825	242677997	242154731	242114397	242114397
chr3	200940782	202683022	198760610	201810706	201097962	201097962
chr4	193253728	191171759	192622708	191569848	191669995	191669995
chr5	186212366	182409348	182259134	181537949	183262120	183262120
chr6	173051588	172871300	172045469	172484233	174742601	174742601
chr7	161716607	161232845	163647842	160431239	160955030	160955030
chr8	145465483	145868755	145680301	145016226	146740677	146740677
chr9	135918162	135032811	137521496	141000299	143803229	143803229
chr10	134385439	135648429	134988382	134604704	135899001	135899001
chr11	133922055	136138868	134265560	135257176	135235771	135235771
chr12	133376810	134127282	133236650	133212390	133580557	133580557
chr13	107377685	112572639	101016140	103308910	113336972	113336972
chr14	102710047	101767573	97561019	95818454	108656801	108656801
chr15	96346030	96851046	95272790	90244609	100881320	100881320
chr16	96215701	87971686	89874169	90313584	90398409	90398409
chr17	86176721	83829559	82991830	82817096	83770182	83770182
chr18	77553665	80564282	76653175	76658275	78908764	78908764
chr19	69461520	61548830	60904190	61496696	61317335	61317335
chr20	66775821	65603693	65674320	66327865	66071306	66071306
chr21	44404722	42759301	38797553	39805135	47311724	47311724
chr22	47079544	48623973	45250002	47559405	53395367	53395367
chrX	155153153	153704768	227216509	154363813	154341397	154341397

I don't see anything suspicious at first sight, other than the translocated duplication of part of chr10 into chrX for RPE1_1 which increases its overall size by ~68%, and was already reported. Could that be the source of all this? Still, it is strange because I tried with even higher values of asm for minimap2 (up to 20) and it didn't work anyway.

[mnshgl0110]

The cigar strings in the error matches the length of chr1 in Hap1 of the three genomes (the first part before S). Based on the sizes (numbers between S and N), I am guessing that these alignments correspond to Chr18, Chr20, and Chr22. You can check the size of the chromosomes in CHM13 to verify that. So, I am guessing that something is weird with the minimap2 alignments.

Probably dotplot would already show that.

Otherwise I would recommend creating a new folder, copy only a single reference and query genome there, and then run minimap2 and syri. This would probably help you figure out whether the tools are actually working as expected.