Optimise Synteny Layout

[Adamtaranto]

I've notice that the macrosynteny plots can often get quite tangled due to arbitrary ordering and orientation differences between the assemblies being compared. Other tools like Mummer4 and Dgenies automate contig reordering and flipping to maximise collinearity.

I think this would be a useful feature, but I don't have time to work on it atm.

If anyone wants to adopt this issue, here is the relevant code from Dgenies as a starting point: genotoul-bioinfo/dgenies#134

[tanghaibao]

There is a layout method but perhaps a bit basic, would be nice to be improved or reimplemented:

python -m jcvi.compara.synteny layout
usage: 
    synteny.py layout query.subject.simple query.seqids subject.seqids

    Compute optimal seqids order in a second genome, based on seqids on one
    genome, given the pairwise blocks in .simple format.
    

options:
  -h, --help   Show this help message and exit
  --qbed QBED  Path to qbed [default: disabled]
  --sbed SBED  Path to sbed [default: disabled]
  --cpus CPUS  Number of CPUs to use, 0=unlimited [default: 10]```

[Adamtaranto]

Cool, I hadn't seen that module. Will leave this issue open to revisit later.

[DougSweeney]

Thanks for showing the layout method, this is helpful. I was also curious if there are any plans to implement the other part of the original posters question, the 'flipping' of chromosomes in the visualization to increase the visual synteny of the chromosomes? Thanks again!

[Adamtaranto]

Hi @DougSweeney, I've implemented a flipping option for the jcvi.graphics.ribbon module and could migrate that into jcvi.graphics.synteny for the microsynteny plots easily enough. Would that be useful for what you are working on?

It would be a bit more work to implement for jcvi.graphics.karyotype to use in the macrosynteny plots.

[Adamtaranto]

@tanghaibao FYI

jcvi/src/jcvi/graphics/ribbon.py

Lines 464 to 469 in bd79635

	def flip(self, gstart, gend, startbp, endbp):
	upstream = gstart - startbp
	downstream = endbp - gend
	newEnd = endbp - upstream
	newStart = startbp + downstream
	return newStart, newEnd

Plus you need to add a field to the layout file to collect the length of each chromosome.

[Lil-Psilocybe]

hello! i'm using jcvi to work with some genomes and noticed the funky swapping for the karyotype (macrosyteny) plots. Has the optimal re-ordering been implemented in macrosynteny plots yet? no worries if not just thought i'd ask.

[Adamtaranto]

Not implemented yet. If your genomes are fairly complete you can manually flip and reorder the contigs in your input fasta file so that they are consistent between assemblies.

[Lil-Psilocybe]

Ah kk. thanks for getting back so quick! and really thankful for this easy-to-use tool