deletions | additions       diff --git a/subsection_FAK_Virtual_screening_results__.tex b/subsection_FAK_Virtual_screening_results__.tex  index b493b42..371ab33 100644  --- a/subsection_FAK_Virtual_screening_results__.tex  +++ b/subsection_FAK_Virtual_screening_results__.tex 
...
\subsection*{FAK} \subsection*{Focal Adhesion Kinase}  Virtual screening results for focal adhesion kinase (FAK) are shown in Figure~\ref{fak}. Similar to ER$\beta$, the only method to outperform random selection is RDKit with an AUC of 0.65. However, in this case FOMS outperforms VAMS with an AUC of 0.48 versus 0.38 indicating that fragment alignment provides some benefit. Also, there is at least one shape constraint query capable of generating an enriched subset with statistical significance. Note that unlike the other kinases, Rho and PKA, for FAK a nitrogen containing ring pattern was not chosen for fragment matching since fewer than half of the actives in the benchmark had such a pattern. Unlike the other kinases, using such a pattern for FAK does not improve the virtual screening results (data not shown).