I would like to use MG-RAST to analyse metagenomic data from paddy soils. I have got the forward (R1) and reverse (R2) fastq files and the file size is aound 20GB (for each R1 and R2). I read some papers and they used denovo assembler first to make the contigs and then uploaded the contigs file. Should I upload the join.fastq files (40GB each) without any denovo assembly (as MG-RAST has Join paired-ends option)? Could anyone please let me know?