Functionality for applying transformations to Jaccard Index of individual SCPs #4
Description
The Jaccard index of each individual SCP may be subject to a transformation on the basis of models stored in the target database. Each of these will be a piecewise linear function of Jaccard similarity (the only value with knots), query and target genome length, and query and target SCP score for the relevant SCP. Likewise, each model will have local, floating point weights to each piece in the function, so the shared SCP count should probable be kept as a double.
In python, this is represented as a function with jaccard, jaccard_slope, intercept, genome_length1, genome_length2, minimum_genome_length_slope, genome_length_diff_slope, score1, score2, mimimum_score_slope, as score_diff_slope, as arguments, but we need a way to make the models available to the parallel processes. If no special models are present, then the unweighted average Jaccard should be taken (which I am representing as a model with slope = 1, intercept = 0, and all other slopes = 0).