Hola Alvaro,

In addition to your approach, I think there are different in silico approaches to evaluate the possible effect of a mutation and to build a hypothesis.

1. I would first start with a group of tools that predict whether a given mutation is pathogenic or benign. Although such prediction tools are not built based on structural modeling, they can be a good start. The following are some links for such tools:

https://loschmidt.chemi.muni.cz/predictsnp/

http://genetics.bwh.harvard.edu/pph2/

2. I would also try to perform some multiple sequence alignment with orthologous proteins to evaluate the degree of conservation of the mutated residue.

I would finally go for the structural prediction tools such as I-TASSER.

Good luck / Buena suerte

Beware of '1-click' solutions.

The purpose of molecular docking, especially when applied to in silico models of proteins, is to generate reasonable protein-ligand complexes and nothing more. The 0.25 kcal/mol is below the sensitivity level of the most successful (luckiest) applications of the best computational approaches.

To assess the effect of the mutation on the ligand binding, quantitatively and qualitatively, run molecular dynamics simulations of the membrane-bound protein-ligand complexes and calculate protein-ligand binding free energies using Free Energy Perturbation (FEP), Linear Interaction Energy (LIE), and/or Linear Response Approximation (LRA) approaches. The sensitivity of these approaches is about 1 kcal/mol.

In the docking calculation, the receptor molecule kept rigid and ligand was flexble. You can't quantify the effect mutant in this way.

You need to peform MD simulation for mutant-ligand complexes and then calculate the free energy using different approaches such as FEP or LIE etc.