appropriate 454-MIDs combination

[alezhe]

Hello
I’ve made a quick search by the thread-subject in the forum and surprisingly found no relevant information.

We want to multiplex some samples for 454 Junior using 48 MIDs.
So question is how to choose appropriate combination of 48 MIDs to minimize the probability of wrong demultiplexing.

I already know that shouldn’t use 16 and 18 together. ))

I can imagine an algorithm for selection but reckon there must be a tool somewhere already.

Thanks.

[martin2]

From my experience you can sometimes see GSMID-like strings in non-barcoded samples. That does not happen much with RLMIDs, if at all. It is probably the increased length of RLMIDs (combinatorial math) what contributes most to eliminate the issue. However, I was never sure whether that also could be because lab space is contaminated by all kinds of oligos, fragmented DNA, etc. and in theory has been being contaminated by GSMIDs for longer time than by RLMIDs.

I think you mean GSMIDs by looking into sequences of "MIDs" 16 and 18. From my previous paragraph it might seem I promote RLMIDs ... no, not really. I would stay with GSMID (TiMID) scheme because with RapidLib protocol you will end up with RLMIDs also on the 3'-end of the read (and will have to remove them unless you do amplicon sequencing and align to a reference!). The only reason for RapidLib protocol is if you do not have enough of sample DNA and you are willing to cleanup your reads afterwards.

Don't use ligase to add MIDs or other oligos, order the adaptor oligos already with the MID included. That is the most important note.