Figure 3: TM-Score Analysis Results on IRF3-R96Q. If our analysis would predict IRF3-R96Q as a cleavage target for 3Cpro, we would expect the TM-Score ratios between IRF3-WT and IRF3-R96Q to follow the same pattern as the controls. Instead, we see "random" ratio values, some above 1 and some below 1, so our analysis does not predict 3Cpro cleavage of IRF3-R96Q.
A t-test between the positive "green" ratios (first 4 columns) and the negative "red" ratios (last 3 columns) results in a p-value of 0.4197, indicating an almost 50% chance of obtaining these results if the two ratio sets were the same.
Our analysis does not predict cleavage of IRF3-R96Q by 3Cpro. Some of the known cleavage targets of 3Cpro were more similar to the known non-cleavage target IRF3-WT than IRF3-R96Q, and some of the known non-cleavage targets of 3Cpro were more similar to IRF3-R96Q than IRF3-WT. This implies the comparison of the structural analysis was essentially random, disallowing a conclusion about cleavage capabilities.
Figure 4: Western Blot shows no preferential cleavage of IRF3-R96Q by 3Cpro in vitro, matching the results predicted by our computational analysis.
Negative results are often not published due to an industry bias against them. This led to difficulty in finding potential cleavage targets that had been shown to NOT be cleaved by 3Cpro, so we had less data points to check.
There were extremely small sample sizes in both the known cleavage target and known cleavage nontarget pools, so the results, even the controlled results, may be pure randomness.
Since we analyzed the difference in structures between SNPs (which, by their nature, have very similar structures), the compared TM-scores were often very close to one another, so each ratio was extremely close to 1, so any small errors may have giant consequences.
The analysis assumes that the sole variable in assessing cleavage capabilities is conformational changes, while other factors, such as changes in amino acid polarity, could play an important factor as well.Â