Generating a novel sequence that does not align with a given genome (Scrabled LNA probe generation)

Question

[gkunz]

This question may be a little out of the norm, but I am trying to synthesize an LNA probe to be used for fluorescent in situ hybridization.

I would like to generate a 22bp sequence that has no alignment greater that 2 nts for a supplied genome file. Even better if there is a way for me to specify GC content of the output sequences. Is anyone aware of a package for accomplishing this or a potentially clever way to go about asking this question?

Thanks for any and all help!

[Matthias Zepper]

I suppose the CRISPR people have developed some tools for designing controls for gRNAs?

If there is nothing helpful and you need to reinvent the wheel, BBTools contains a Bloom-Filter implementation, so it can separate reads that certainly do not match a genome reference (you however will need to adjust the parameters for the 2nts criterion). For 22-mers, it would be quite feasible to do the brute force approach of designing all possible sequences and running them through the filter, although that will most likely give you a few thousand possible sequences, which you could at least narrow down by the desired GC content. Lastly, you would need to check for hairpin-formation and maybe some other properties?