Hi there, Following my yesterday's message, I changed my original idea and finally parsed the mira-produced caf file to obtain a phd.ball file to be used with consed. The idea behind that is to have qualities associated with reads when editing mira assemblies within consed. This is very important for example when merging/tearing contigs, because the consensus is recalculated in a very, very bad way if you don't have qualities (especially because mira doesn't physically trims the reads from the vector sequences...). The result is a small perl script that works for my data, but I would be glad if others could test it to see if it works with other types of data. All comments are welcome! CAVEAT: this script produces huuuuge files, because it writes one line per base, plus headers. For example, I have 350'000 reads and some long Sanger, and I get a file which is 1.4 Gb... Cheers, Lionel