HI,

I would first assemble the reads using velvet or SOAPdenovo and then use MEGAN to see the %age of contigs mapping to different genomes after blast alignments. Then there are many nice assemblers for metagenomic studies: MetaVelvet, Met-AMOS, MAP (http://bioinfo.ctb.pku.edu.cn/MAP/). Please go through the literature regarding it, you will find many articles showing performance and benchmarking of these tools.

Best, Rahul

See this Assembly Illumina Paired End Reads. I assume you're just "pairing" your mate pairs and not assembling the metagenomic data into larger contigs. I think you should initially use individual reads; I would avoid assembling your metagenomic data into contigs until you get a better idea what organisms your data represents. You'll next want to identify them: using BLAST or any other number of platforms for metagenomic analysis. "Mapping" in a transcriptome sense doesn't work well for metagenomic data.

This paper might be of benefit to you: Assembling large, complex environmental metagenomes

They don't really incorporate mate-paired data, so I think it might be best to use some of their grouping/pre-assembly step and then switch over to ALLPATHS-LG, which uses short Illumina reads to generate contigs and then uses mate-paired data to group the assembled contigs and put them together.

If I were you, I would start with this pipeline: Differential gene and transcript expression analysis of RNA-seq experiments with TopHat and Cufflinks. Especially this bowtie-tophat-cufflinks pipeline is very apt for illumina sequencing data in general.

Thanks for all the suggestion. yes it is whole-genome "shotgun" metagenomic data from Illumina with 101bp paired reads. i have three libraries 1. 180bp overlapped paired library, 2. 2K mate-pair library, 3. 5K mate-pair library

I appreciate your suggestion

Thank you