QSARs for 6-azasteroids as inhibitors of human type 1 5 alpha-reductase: Prediction of binding affinity and selectivity relative to 3-BHSD

Quantitative structure-activity relationships (QSARs) are developed to desc ribe the ability of 6-azasteroids to inhibit human type I 5 alpha -reductas e. Models are generated using a set of 93 compounds with known binding affi nities (K-i) to 5 alpha -reductase and 3 beta -hydroxy-Delta (5)-steroid de hydrogenase/3-keto-Delta (5)-steroid isomerase (3-BHSD). QSARs are generate d to predict K-i values for inhibitors of 5 alpha -reductase and to predict selectivity (S-i) of compound binding to 3-BHSD relative to 5 alpha -reduc tase. Log(K-i) values range from -0.70 log units to 4.69 log units, and log (S-i) values range from -3.00 log units to 3.84 log units. Topological, geo metric, electronic, and polar surface descriptors are used to encode molecu lar structure. In formation-rich subsets of descriptors are identified usin g evolutionary optimization procedures. Predictive models are generated usi ng linear regression, computational neural networks (CNNs), principal compo nents regression, and partial least squares. Compounds in an external predi ction set are used for model validation. A 10-3-1 CNN is developed for pred iction of binding affinity to 5 alpha -reductase that produces root-mean-sq uare error (RMSE) of 0.293 log units (R-2 = 0.97) for compounds in the exte rnal prediction set. Additionally, an 8-3-1 CNN is generated for prediction of inhibitor selectivity that produces RMSE = 0.513 log units (R-2 = 0.89) for the external prediction set. Models are further validated through Mont e Carlo experiments in which models are generated after dependent variable values have been scrambled.