deletions | additions       diff --git a/untitled.md b/untitled.md  index 1fbe1f0..31a088d 100644  --- a/untitled.md  +++ b/untitled.md 
...
Previous efforts to computationally model the relationship between the thermal stability of enzymes and point mutations in enzyme systems have relied on the addition of evolutionary information (Thompson) and, see CASP, have had success in creating models of proteins to an accuracy of X previously-believed to be impossible to model accurately. People have tried to model the biophysical constraints on protein evolution but they rarely have any actual mutational data to draw from (only the "fossil record" of known holomologus sequences and the evolutionary history of the organisms that we see a snapshot of) \cite{25165599}  Here, we measure the thermal denaturation temperature of 140 123  mutants of a family 1 glycoside hydrolase, BlgB. After generating models of each mutant, we combine the experimental data with 45 features (e.g., total system energy, ligand energy). A machine learning algorithm trained on the data is used to make a prediction for the thermal stability of 15 point mutants evenly sampled from the single nucleotide polymorphism--accessible space and 15 point mutants evenly sampled from the elastic net model. The predictive model achieved a Pearson correlation coefficient of [...] in our tests, showing that [...]