1

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.

chubs
  • 33
  • It depends on your theory, because drugs can be synergetic or antagonistic. It also depends on the biological meaning of this numeric result, as well as how it was measured. We need more details of your circumstances to be able to answer this question. – whuber Jan 13 '23 at 13:48
  • Sorry for not clarifying. I used drugs and outcomes as an example (probably a bad one). In reality these are computer-generated predictions ("outcomes") with different perturbations ("drungs"). Therefore, the outcome of a single perturbation usually doesn't have an effect, whereas having multiple perturbations sensitizes the system to have a more dramatic effect on the outcome. I have run all possible combinations of perturbations and would like to extract which of the perturbations had the strongest effect. – chubs Jan 13 '23 at 13:54
  • 1
    Please edit your post to describe the situation you really have. This formulation is so general it requires a textbook or more to answer. – whuber Jan 13 '23 at 13:56
  • 1
    I've modified the question – chubs Jan 13 '23 at 14:00
  • I've been trying to write an answer, but it's hard as you haven't specified just what the "Outcome" is and how it compares to the un-mutated form of the protein. Is this perhaps some estimate of enzymatic activity relative to un-mutated having a value of 1? For "sensitization" do you mean increasing or decreasing the value of outcome overall due to presence of a type of mutation, or if a particular mutation magnifies the magnitude of a deviation from, say, a normal value of 1 that was associated with other mutations? Please clarify by editing the question, as comments are easy to overlook. – EdM Jan 14 '23 at 23:02

0 Answers0