Right now, EMIRGE uses bowtie for its read mapping, which does not handle indels. Consequently, EMIRGE is not coded to handle indels either. To be able to get more accurate reconstructions, as well as use homopolymer-rich 454/Roche sequencing data, we need to incorporate indels into the mapping and statistical model.
Right now, EMIRGE uses bowtie for its read mapping, which does not handle indels. Consequently, EMIRGE is not coded to handle indels either. To be able to get more accurate reconstructions, as well as use homopolymer-rich 454/Roche sequencing data, we need to incorporate indels into the mapping and statistical model.