Repeat sequences-Genome assembly

[bioman1]

I have WGS sequence of illumina paired-end reads (2x101 bp) of the plant crop. I have de novo assembled the reads using velvet.

I would like to do two things

1. To do de novo genome assembly and identify repetitive sequences. I know i can do with repeat masker, is there any software other than this?
2.Remove prior repetitive sequence from illumina reads and do de novo assembly.

Please let me know software to find repetitive sequence before assembly and after assembly.

[Brian Bushnell]

As I mentioned in this thread, you can use a tool such as BBNorm to get a kmer frequency spectrum of your reads, and look at the peak locations to determine the coverage range for 1-copy, 2-copy, etc peaks. Then you can bin them by kmer frequency. Illumina coverage is highly uneven so this does not work perfectly. And of course it's harder with ploidy over 1, but still possible... to some extent. Then you can just assemble the reads in the unique bin.

The repeat stuff may not assemble very well with Velvet. But another way to do it is by mapping the input reads to your assembly, and annotate/sort the contigs by coverage; contigs with 2x median coverage may be 2-copy repeats, and so forth. Velvet already annotates contig names with their estimated coverage, so you could use that instead. Either way, once you isolate the repeat contigs and unique contigs, separate them into two fasta files and map the reads to them with BBSplit to get three outputs - a pile of single-copy reads, a pile of repeat reads, and a pile of unmapped reads.

[ssully]

the introduction of this paper contains nice brief descriptions of many of the softwares used to find and analyse genomic repeats.