Skip to main content
. 2022 Nov 7;61(22):2461–2469. doi: 10.1021/acs.biochem.2c00499

Figure 5.

Figure 5

Correlation between experimental (Δetox) and predicted (Δptox) changes in cytotoxicity upon TDP-43 single (A) and double (B) missense mutants. Predictions were performed with the extended FuzDrop method based on three parameters: the change in droplet-promoting probability (ΔpDP), amyloid-promoting probability (ΔpAP), and change in multiplicity of binding modes (ΔMBM). (A) Prediction of the cytotoxicity of single mutants. Application of random forest models (Methods) on 498 single variants. (B) Prediction of the cytotoxicity of double mutants. Application of random forest models (Methods) on 23,802 double variants, where the parameters were averaged for the region of residues 312–341. Only 10% of the data is shown for clarity, and the R value is computed for the whole data set (Table 1). Pearson’s correlation coefficients were calculated in R. The cytotoxicity scale ranges from green (not toxic) to red (toxic) and is shown on the right panels. Cytotoxicity values were taken from ref (19).