Is there any verdict on using the iAssembler tool? I would assume you can simply tweak mira's parameters to get more things to assemble together rather than run mira 4 times in a row. On Wed, Feb 16, 2011 at 1:59 PM, Sven Klages <sir.svencelot@xxxxxxxxxxxxxx>wrote: > 2011/2/16 Jeremy Volkening <volkening@xxxxxxxxxxxxx> > >> On Wed, 2011-02-16 at 21:19 +0100, Bastien Chevreux wrote: >> >> > Now on to why I think this kind of iterational assembly is a bad idea: >> > you lose accuracy ... a lot. Consider a case where you perhaps have a >> > ploidy variants with one column having 50 T's and 50 A's and one C >> > (the A and T being real, let's a ssume the C is a sequencing error). >> > > Just to give my 2p, .. though it is slightly off-topic. People are asking > me the same questions as the OP, they want less contigs and care about the > debris / singleton stuff with medium sized 454 EST datasets (300,000 - > 1,000,000 reads). Depending of the library and the dataset, there are some > 50%-90% of the reads assembled into contigs. They do care about the rest ... > thus asking if there is a way to get these reads assembled too (some check > with blast if the reads hit contigs). If then the reads are blasted against > nrprot or something similar, it is not really important if there is an "A" > or an "T" at that position for characterising the potential transcript ... > that is the "usual" argumentation. > > So people tend to re-assemble the assemblies ... this is one "re-Assembler" > using MIRA and cap3 (I have not tested this package), > http://bioinfo.bti.cornell.edu/tool/iAssembler/ > > cheers, > Sven > -- Artemus Harper