My edits to the paper. These are mostly minor, but one more significant change is I removed most reference to "fitness" or "fitness landscape", replacing with "function/phenotype" and "genotype-phenotype map". Protein researchers often equate protein fitness to function, so this is sort of a normal thing to do, but it is pretty confusing: proteins don't have fitness in an evolutionary sense, an organism (virus) has a fitness that depends on protein function. Since we are also interested in modeling actual evolutionary fitness as a function of protein function (e.g. in GCreplay), this is especially problematic. The flow is genotype-->phenotype-->fitness, and torchdms is only about the first link.
Feel free to accept/reject changes as you see fit.
My edits to the paper. These are mostly minor, but one more significant change is I removed most reference to "fitness" or "fitness landscape", replacing with "function/phenotype" and "genotype-phenotype map". Protein researchers often equate protein fitness to function, so this is sort of a normal thing to do, but it is pretty confusing: proteins don't have fitness in an evolutionary sense, an organism (virus) has a fitness that depends on protein function. Since we are also interested in modeling actual evolutionary fitness as a function of protein function (e.g. in GCreplay), this is especially problematic. The flow is genotype-->phenotype-->fitness, and torchdms is only about the first link.
Feel free to accept/reject changes as you see fit.