Hello All, When I read this http://thegenomefactory.blogspot.in/2012/09/using-velvet-with-mate-pair-sequences.html ( by Torsten Seemann ), i came across Combining all of them(mate-Paired end, Paired-end and Single End reads) for denovo assembly. I would like to know that why all 3 libraries (SE reads, PE reads and Mate PE reads) of an organism should be combined and assembled? Combining 3 libraries will improve the assembly quality?
That blog post is not a recommendation, it's an example. Just because you can do something, doesn't mean you should. In this case the recommendation is about specifying the input order for Velvet - i.e. smallest insert size first. Generally for a short-read de novo assembly you would use paired-end and mate-pair data only. Adding mate-pair data is definitely a great way of improving an assembly that has been generated from paired-end data.
Mater-pair information has a high level of duplication. And the reason you need mate-pair is only for arranging fragmented blocks based on distances (imagine mile markers). If you need to actually assemble data, you need information based on adjacencies, that is what you get from paired-end information. Even though you get blocks from adjacencies, they are broken either due to repeat content or genome complexity or lower coverage regions. This is where our mile-markers of mate-pairs come in for arranging continuous blocks together.
Yes you need to combine both for more information as paired-end gives you fragmented blocks and mate-pair the distance estimates, which need to be put together.