When I last looked at Julia (as a candidate for writing a bioinfo. tool) it seemed a really appealing alternative to prototyping first in Python and then Cythonizing to gain speed, in that code would only have to be written once and no stitching together would be required. Ultimately, I decided against using it because it seemed packaging of binaries wasn't quite there yet (e.g., one would have to download Julia and then download my code to run it). Is this something that's likely to be addressed in 0.4? Was my understanding of the situation accurate?
I have nothing against service in the armed forces, but I doubt it would be more lucrative. You may sign up to be a pilot and wind up doing something completely different (if you don't make it through the training). Also, if consider time on base as time on the job, it may equate to less than minimum wage overall. At least, that's the way it is with mandatory service in Israel. Not sure how it works in the US though.
I am also a bioinformatician. How much control would you have over node resources? What happens if a browser goes from idle to active? Also, calling this sequencing isn't accurate - this is alignment as daemonk pointed out.
You pick words you don't know, it makes flashcards and suggests stuff at your reading level - seems like it would be awesome for reading (academic) papers.
Can you explain "probabilistic key-value store?" Would it be that each gene has some defined probability of belonging to a given organism, or is it probabilistic in the sense of having a defined error rate as BFs do?
You could use them for rough spell-checking by hashing all the words in a dictionary. You would then need some additional data structure for overcoming false positives though, but any word rejected would necessarily not be a member of the original dictionary. The BF size would be much cheaper to store than the original dictionary and searching for each word would be pretty fast (though not necessarily faster than exact string lookups).