merging of two assemblies from different short-reads techniques

Question

[Andrew Yu]

Hi all!

I have two data sets for one metagenome sample:

1. Illumina 2x300 bp shotgun metagenome library.
2. DNBSEQ 2x100 bp shotgun metagenome library.

THe question is: is it possible to get more deep and complex assembly by combining those datasets before\after assembly step?

I've tried just merge forward and reverse reads from both Illumina and DNBSEQ into mergred_forward.fastq and mergred_reverse.fastq and pass it to SPAdes or megahit. But I feel this is wrong approach. I figured out that genes of 16S rRNA look more accurate from separated assemblies than from merged.

Also I've heard about unicycler but it needs long-read sequencing data which i haven't. also I thought about using assembled contigs of Illumina or DNBSEQ data as pseudo-long-reads and pass to unicycler but it seems illegal.

So I'm cunfused and need some advice at this point.

Thanks a lot!

[GenoMax]

I've tried just merge forward and reverse reads from both Illumina and DNBSEQ into mergred_forward.fastq and mergred_reverse.fastq

Can you clarify what you mean by this?

Did you just cat the forward read files into one? I think that is what you seem to have done based on names of files.

OR

Did you try to merge individual reads from each of the technologies (e.g. using a tool like bbmerge.sh referred to below) into a single long read (assuming reads are overlapping i.e the length of sequencing is > than size of insert)?

[Andrew Yu]

Hi! Yes , I just use cat for iilumina fiorward and dnbseq forward and same for reverse and pass in to assembler. I haven't tried yet bbmerge.sh so I'll try it right now. If I am clearly understand this tool will just mate my paired reads into single-end reads?

[GenoMax]

If I am clearly understand this tool will just mate my paired reads into single-end reads?

It will merge the overlapping part of the reads into one long read, only IF the two reads overlap. If they don't then nothing will happen.

R1  -------------------------------------->
    ----------------------------------------------------------------------------- Lib fragment
                                 <---------------------------------------------------  R2

Answer 1

[Vijith]

You may read about bbmerge here: https://github.com/BioInfoTools/BBMap/blob/master/sh/bbmerge.sh, then you can try to them using SPAdes.

[Andrew Yu]

Is it OK for bbmerge and SPAdes that my reads have diffenet length? i.e. Illumina is 2x300bp and dnbseq is 2x100bp.

[Vijith]

yes, I think the tool is quite adaptable and can work with reads of different lengths.