Newbe struggling

[dhatziioanou]

Hello everyone,

I'm new to genome assemblies and am trying to patch together de novo a genome without an available reference (expected size 20-100 Mb).
I've got about 32 files of paired end Illumina MiSeq reads with mostly 150bp but also 75bp reads which I've been trimming with cutadapt, fastqGroomed, currently trying to make Quake work on them and hopefully that will help stitch it all together in a better way than my employer already has which results in over 66.000 scaffolds! For the assembly I'm going to try SOAP de novo and ALLPATHS-LG.

Any tips or links to pipelines for something similar would be appreciated. Also can anyone tell me if I should be combining my sequence files before running Quake or not? FastQC shows me I've got between 8 and 35% duplicates in my sequence files.

[GenoMax]

Hello everyone,

I'm new to genome assemblies and am trying to patch together de novo a genome without an available reference (expected size 20-100 Mb).

That is a rather wide range. Is that the best guesstimate you have? Is there a related genome available that can allow you to refine that range? Is this a haploid, diploid or "something else" genome? Is anything known about presence/absence of repeats?

[dhatziioanou]

So far the only sequenced genomes of closely related species I can find are 27-32 Mb but other slightly further related species range from 13-100 Mb.
It should be a diploid and I'm expecting plenty of repeats and HGT.

Basically I'm thinking that using Quake and then khmer may help selectively assemble the best sequences into a backbone to then assemble the rest of the sequences around, any opinions on this? I technically have well over 50x coverage from the reads so I'm thinking the coverage should theoretically be good enough to sacrifice as many sequences necessary in the name of quality and a more realistic assembly.

[westerman]

You could try the 'mira' assembler which should handle a lot of the potential problems and can work on smaller genomes.