Transcriptome assembly improvement with Soap-denovo trans

Question

[seta]

Hi all,

I'm trying to assemble about 300 million PE Illumina reads (100bp) using Soap-denovo trans assembly with kmer of 27, which generated too many scaffolds (about 355000, min length of 100), the contig and scaffold numbers are so close together, sounds that scaffolding has not been well done. Could you please share me your idea to improve the assembly, which parameter(s) is more important on this issue?

Thanks in advance