running ARCS multiple times using newly scaffolded assembly as a reference about arcs HOT 6 CLOSED

Thank you for sharing your results, we are grateful to receive this feedback and see how useful ARCS/LINKS has been in your project.

It is sensible to use it iteratively;

We have done limited iterative tests in the past, and our starting assembly was quite contiguous perhaps it is why we did not observe appreciable gains after the 2nd iteration. It makes sense that you would be making merges after several iterations, as the scaffold graph becomes simpler and simpler, with less conflicts. I recommend trying another set of iterative scaffolding, but with more stringent (-c 4/5, -l 5, -a 0.3/0.5/0.7) parameters this time and compare both resulting assemblies for completeness (with BUSCO).
Cheers,
Rene

from arcs.

Thank you, Rene, for quick reply!
We will try running ARCS+LINKS with more stringent parameters.
Best regards,
Mikhail

from arcs.

Hi Rene,

I have tried more stringent parameters and, unfortunately, after examining the scaffolds after ARCS-LINKS pipeline, I realized that the scaffolds that I've obtained were chimeric. I've plotted the start and end of each barcode (>10) in new scaffolds to visualize the pattern.

The parameters that I've used for ARCS:
s 98 -c 5 -l 0 -m 25-20000 -d 0 -r 0.05 -e 30000 -z 400
for LINKS:
-l 5 -t 2 -a 0.7 -x 1 -z 400

Here are some examples:

On the other hand, it seems that there is lot of information in the data.
Is there a way to get our assembly more continuous, and keeping errors at low level?
Playing with -e parameter? Something else?

best regards,
Mikhail

from arcs.

Thank you for the insights Mikhail.
The question is, were they misassembled to begin with? We are actively developing a correction utility to help catch these and rectify assemblies (under review). Because ARCS looks at -e bases for scaffolding, and the ends might be misassembled to begin with, I recommend augmenting -e (we have done limited tests with -e 60000). If you suspect that the misassemblies you are seeing are induced by ARCS, I recommend increasing -l (10, 15, 20..) and decreasing -a (0.1, 0.2, 0.3, ..).
Thank you for sharing,
Rene

from arcs.

Thank you,

I'll try more stringent parameters.
I suppose that misscafolding most likely was introduced after ARCS in our case.
Here are the examples of barcode distribution in some original longest scaffolds from the CLC assembler:

Best regards,
Mikhail

from arcs.

Closing this old issue -- feel free to re-open if you still have questions.

from arcs.