I have generated structure predictions of a protein-protein interaction with different mutations at the interface. The outcome of a single mutation usually doesn't have an effect in this system, but having multiple mutations sensitizes it to have a more dramatic effect on the interaction. I have run all possible combinations of mutations (labelled A to D below) and would like to extract which of them had the strongest effect. It is clear that the presence of some of the mutations is more likely to sensitize than others.
| Mutation | Outcome |
|---|---|
| A | 1 |
| A_B | 0.3 |
| A_C | 0.2 |
| A_D | 0.4 |
| B | 0.2 |
| B_C | 0.4 |
| B_D | 0.1 |
| C | 0.5 |
| C_D | 0.5 |
| D | 0.4 |
| A_B_C | 1 |
| A_B_D | 1.2 |
| etc. |
What analysis is most appropriate to find the contribution of each mutation (e.g. A) to the outcome? Is this a problem for PCA? I am coding in python and have organized my data into a dataframe.