GFA `A` lines about hifiasm HOT 9 CLOSED

Here is an example of A line:

A	utg000001l	2093	-	SRR11606870.1250244	0	16611	id:i:1250243	HG:A:a
A	utg000001l	3572	-	SRR11606870.2648803	0	17169	id:i:2648802	HG:A:a

For SRR11606870.1250244, 'utg000001l' is the unitig name, 2093 and (3572 - 1) is the start position and end position in utg000001l, '-' is the direction in utg000001l, 'HG:A:a' is the haplotype label of SRR11606870.1250244 which is only useful for haplotype-resolved assembly like trio-binning.

As for cigar, I personally think most reads are exactly mapped to unitig since all reads have been corrected. I'm not sure if this assumption is enough for your project...

from hifiasm.

Thanks for the quick response - just clarifying, there are 2 A lines above. It looks like they refer to different reads. I assumed that 2093 and 3572 are the start coordinates of alignment on the unitig and 0 is the start coordinate on the read in each line, and 16611 and 17169 are either the end coordinates or the alignment lengths?

I agree the reads should match the contigs almost exactly - for now it is useful enough for me to just to pull the coordinates, although if it would be easy to put in a cigar, or even better paf-style cs tag that would be really helpful, since I assume there will be some bubbles that are popped, correct? Or would the popped reads just get discarded?

from hifiasm.

It is technically difficult to output CIGAR because a small fraction of reads are not mapped exactly. We have no plan to deal with those. In addition, everything on the A-line is derived from corrected, not raw reads. CIGARs from corrected reads won't be useful to you anyway.

Or would the popped reads just get discarded?

Popped reads and contained reads are discarded. If you want to recruit reads in a region, you have to redo read alignment against the contigs.

from hifiasm.

Ok, thank you anyway!

from hifiasm.

I agree. Another possible solution is to jointly check p_ctg, a_ctg and r_utg. It will be more accurate in extreme cases, but maybe not as easy as directly alignment.

It is technically difficult to output CIGAR because a small fraction of reads are not mapped exactly. We have no plan to deal with those. In addition, everything on the A-line is derived from corrected, not raw reads. CIGARs from corrected reads won't be useful to you anyway.

Or would the popped reads just get discarded?

Popped reads and contained reads are discarded. If you want to recruit reads in a region, you have to redo read alignment against the contigs.

from hifiasm.

I assumed that 2093 and 3572 are the start coordinates of alignment on the unitig

Yes

and 0 is the start coordinate on the read in each line and 16611 and 17169 are either the end coordinates or the alignment lengths?

No, 16611 or 17169 is just read length, instead of alignment length.

from hifiasm.

Hi,
I wonder what the number 1250243 after the "id:i:" in "id:i:1250243" means. thanks!

from hifiasm.

It is the read ID, which is only useful for hifiasm itself.

from hifiasm.

i see. thanks!

from hifiasm.