Researchers using Velvet for genome assembly encounter the problem of which k-mer size to use to get the best genome assembly. Velvet Optimizer is a freely available multi-threaded perl script that runs multiple velveth and velvetg steps to find optimal Velvet parameters for a library of reads. It was originally written by Simon Gladman of Commonwealth Scientific and Industrial Research Organisation in Australia.
The commands to run ‘velvetoptimizer’ are explaind in this thread. If we find other links, they will be reported here.
Tsuyoshi Hachiya and his colleagues from Keio university wrote a modified version of Velvet for assembling metagenomes. They used a partitioning step to partition the de Bruijn graph of all reads, and then assembled each partition independently.
We previously commented that metagenomes and transcriptomes have many similarities. Conceptually, their approach indeed appears similar to Trinity, but I have not checked the details.
We are going through the Velvet and Oases mailing lists of the last three years to find useful ideas for genome/transcriptome assembly. Several very smart and experienced researchers comment there from time to time. We plan to use these Velvet and Oases threads of our forum to report what we find, and once done, we plan to write a full blog post summarizing all ideas. The mailing lists have almost thousand posts and it is taking us some time to go through them all. If you like, please feel free to add.