3d-dna creates large contigs that consist of multiple chromosomes

[Benedikt]

Hello,

I am trying to use juicer and 3d-dna to polish a FALCON assembly. We sequenced a different strain of a known reference organism and observed large amounts of chromosomal fusions introduced by juicer and 3d-dna. To avoid a technical artifact, I assembled the reference strain as well and here I see the same thing. We know there should be 11 chromosomes, but 3d-dna produces 2-3 mega scaffolds that contain parts from multiple chromosomes and a lot of smaller fragments. It seems that the "split" step is runnic erratic. Is there any guideline on troubleshooting or parameters that could be changed? 

Thank you for your time.

[Olga Dudchenko]

Hi Benedikt,

Does .0.hic look fine? Can you share the screenshot with .wig and .bed tracks from the 0 step loaded on top?

Olga

[Benedikt]

Hi Olga,

apologies but I am not sure what you mean. I looked at all the hic files in juicebox, but I don't see anything peculiar. I am attaching the svg from the .0.hic. 

How can one load .wig and .bed files on top? 

Thank you very much for your help!

[Olga Dudchenko]

Hello Benedikt,

Well, I would venture to disagree and point that your map is extremely peculiar. You sample has huge coverage biases and is karyotypically abnormal. You will not be able to use default 3d-dna with this assembly. I am not sure if one can speak meaningfully of an assembly here at all, but if you want to play around with the data and better understand what you've got there, I recommend downloading the .0.hic and .0.assembly and playing with them in JBAT. For more on JBAT see Genome Assembly Cookbook on dnazoo.org/methods.

Best,

Olga