fraizerangus,
It sounds like you have a lot of "lines" and a single reference "point" that you want to calculate distances on. You have been provided a number of good options.

I was a bit curious as to a PDB file so I googled it and found this and this which suggest that you are provided with an X,Y and Z of the atom in relationship to the cell's origin. This is just a point, not a line. Where are your "lines" coming from?

I was wondering how many computations needed to be made, what accuracy was required and how important runtime was. For instance, would using Inline::C be in order? Would it be possible to take advantage of a loose margin of error to normalize the inputs to use a cache lookup (either precomputed or memoized) for speedier results. None of this is actually important - it just seemed like a potentially interesting problem so respond if you have time.