Getting nuclear genomic reads for denovo assembly-HELP

[bioman1]

I have plant genomic reads (illumina paired-end reads, Hiseq 2000 -WGS approach). I would like to get nuclear genome eliminating probable contaminating sequences (like mitochondrial, chloroplast, bacterial sequences and vector sequences). After eliminating, I would like to do denovo assembly.

I would like to know the good workflow for this. Please help me in suggesting good workflow. The planned workflow

1. I am planning to map the filtered illumina paired-ends reads (filtered through trimmomatic tool) to Arabidopsis thaliana mitochondira & choloroplast genome uisng BWA and filter unmapped reads using samtools.

2. The unmapped reads will be nuclear reads.Again vector and bacterial contamination is removed by mapping against univec database (http://www.ncbi.nlm.nih.gov/tools/vecscreen/univec/) and bacterial genomes (http://www.ncbi.nlm.nih.gov/genomes/...l_taxtree.html).

My question is , how can I map the reads

(i) To i need to index reference genome individually or can I combine (chloroplast and mitochondira genome as one reference)?. If I need to index separately, how can I get unmapped reads from choloroplast and mitochondria genome as paired-end fastq file?
(ii) After getting unmapped reads, how can I remove bacterial and vector contamination from the reads?

[Brian Bushnell]

(i) To i need to index reference genome individually or can I combine (chloroplast and mitochondira genome as one reference)?

Combining them is most efficient and has no drawbacks.

(ii) After getting unmapped reads, how can I remove bacterial and vector contamination from the reads?

Well, by mapping, again. But just roll the vector, bacterial, mito, and chloro reference into one fasta and do all the mapping at once.

If you have a closely-related plant, a better approach would be to use BBSplit, and then use both the reads that map to the related plant and the unmapped reads, but throw away the reads that map to the other things. That will prevent false positives - removal of reads that actually do belong to the plant - which would cause fragmentation of your assembly. If you don't have a close relative, then don't worry about it.