This is a slightly odd problem but I am hoping someone in the audience has encountered this. I have a large library on shRNA's on about 350 96 well plates. Each well has a unique shRNA. About 3 to 4 different shRNAs target a given gene ,but these 3-4 shRNA's can be anywhere in the 350 plates. I would like to reduce the size of the library by pooling shRNA's that target the same gene. That is make a new pooled library , by picking from the existing library (keeping the original library intact). Constraints: I have to use a robot, that can handle only 10 plates at a time. (one (or more?) of the positions will have to be the used for plates I am moving the genes to) I can choose any 10 plates I like. Say I only wanted to do this 10-plate-shuffle 20 times. Note robot moves are REALLY SLOW ! What are the optimal 10 plate choices I can make to reduce the library to the smallest number of plates (making shRNA gene specific pools)? I was going to build a matrix of "gene- overlaps" on each plate and look for best overlaps , recalculating after each pooling... then clean up partial plates afterwards. Perhaps someone has some better optimization algorithms or have done this already? Thanks in advance Paul [[alternative HTML version deleted]]