plasmid assemby from Illumina reads - help!

Question

[rebeccarenberg]

I have some plasmids (purified) that were 1x150 sequenced on Illumnia. I am trying to assemble them, but am not having success using SPAdes or tadpole.sh from BBTools (both have been suggested by others to assemble viral genomes which are comparable in size). I have plenty of coverage with the number of reads generated (1000+X coverage). SPAdes produces 1000 contigs with the longest one being 1,482 bases. tadpole.sh produced 19 contigs with the longest being 912 bases.

I feel like I am missing some parameters somewhere to get a better assembly.

Here is the SPAdes command:

spades.py --careful -s XXXX.fastq.gz -t 24 -m 31 -o SPAdes_assembly_XXXX

Here is the tadpole.sh command:

tadpole.sh in=XXXX.fastq.gz out=contigs_XXXX.fa k=93

Any suggestions on what parameters I should change to get a better assembly (i.e the full length plasmid in a single contig)?

Is there a special tool out there to do plasmid assemblies (I tried plasmidSPAdes.py too with no luck).

Any other information that would be helpful?

Thanks!

Answer 1

[GenoMax]

1x150

Is that accurate? Or was it a paired-end run and was actually 2x150?

You may want to normalize read coverage (if you have a lot of it, which is known to cause problems with assembly). Take a look at tadpole.sh guide here and bbnorm.sh guide here.

[rebeccarenberg]

Yes, the 1x150 sequencing is accurate. Thank you for the suggestions.

I normalized using

bbnorm.sh in=XXXX.fastq.gz out=XXXX_normalized.fq target=100 min=5

then I ran the assembly again using

tadpole.sh in=XXXX_normalized.fq out=contigs_XXXX_normalized.fq k=93

This assembly was 19 contigs, with the longest being 1002 bases. Still not what I am hoping for. Any other suggestions?

[rebeccarenberg]

target=100 was too high. I dropped it down to 20/30/50 and I got a beautiful assembly. THANK YOU!!! I would have never figured out how to do the normalization or that I needed to.

[GenoMax]

Excellent. I moved my comment to an answer. Feel free to accept it to provide closure for this thread.

Answer 2

[rebeccarenberg]

Recently I discovered Unicycler (https://github.com/rrwick/Unicycler ), which worked really well to assemble these plasmids and I didn't have to normalize the reads first (win - one less step).