Use of Stratified Cascade Learning to predict hospitalization risk with only socioeconomic factors.

BACKGROUND AND OBJECTIVE: Published models predicting health related outcomes rely on clinical, claims and social determinants of health (SDH) data. Addressing the challenge of predicting with only SDH we developed a novel framework termed Stratified Cascade Learning (SCL) and used it for predicting the risk of hospitalization (ROH). MATERIALS AND METHODS: The variable set includes 27 SDH and "age" and "sex" for a cohort of diabetic patients. The SCL model uses three sub-models: SM1 (whole training set) stratifies training set into "predictable" and "unpredictable" subsets, SM2 (built on whole training set) classifies test set patients into "predictable" and "unpredictable", and SM3 (built on only the "predictable" subset) predicts the ROH for the patients classified as "predictable" by SM2. RESULTS: The SCL model does not improve either the AUC or the NPV of the basic classifier, but materially improves accuracy and specificity measures at the expense of lowering sensitivity for the "predictable" subset. Optimization of the risk thresholds of the sub-models does not noticeably change the AUC and NPV but further improves the accuracy and specificity at the expense of further lowering sensitivity. CONCLUSION: Since the SLC model yields low sensitivity it fails to predict high risk patients. But it yields high specificity that can be useful when the objective is to eliminate low-risk patients as candidates for further testing or treatment. The use of the SCL is not limited to healthcare, it can be applied to any predictive modeling problem when reliable predictions can only be made for a fraction of incoming data.

Synthesis and structure-activity relationship studies of novel dihydropyridones as androgen receptor modulators.

A library of 3-hydroxy-2,3-dihydropyridones was synthesized, and their activities as antiandrogens were tested in the human prostate cancer cell line LNCaP. Structure-activity relationship (SAR) studies resulted in the identification of a potent compound whose activity is comparable to that of MDV3100. Homology modeling and molecular mechanics were used to build a structural model of the androgen receptor-ligand binding domain and to investigate the structural basis of the antagonism. The model is qualitatively consistent with the observed SAR. Moreover, the enrichment plot shows that screening with the model performs significantly better than random screening. Therefore, the model probably represents a realistic conformation of the antagonist form and can be utilized for structure-based design of novel antiandrogens.

Novel Nitrobenzazolo[3,2-a]quinolinium Salts Induce Cell Death through a Mechanism Involving DNA Damage, Cell Cycle Changes, and Mitochondrial Permeabilization.

This study reports the capacity of three nitro substituted benzazolo[3,2-a]quinolinium salts NBQs: NBQ 95 (NSC-763304), NBQ 38 (NSC 763305), and NBQ 97 (NSC-763306) as potential antitumor agents. NBQ's are unnatural alkaloids possessing a positive charge that could facilitate interaction with cell organelles. The anticancer activities of these compounds were evaluated through the National Cancer Institute (NCI) 60 cell line screening which represents diverse histologies. The screening was performed at 10 µM on all cell lines. Results from the NCI screening indicated cytotoxicity activity on six cell lines. In order to explore a possible mechanism of action, a detailed biological activity study of NBQ 95 and NBQ 38 was performed on A431 human epidermoid carcinoma cells to determine an apoptotic pathway involving, cell cycle changes, DNA fragmentation, mutations, mitochondrial membrane permeabilization and caspases activation. DNA fragmentation, cell cycle effects, mutagenesis, mitochondrial permeabilization and activation of caspases were determined by fluorimetry and differential imaging. Our data showed that A431 growth was inhibited with an average IC50 of 30 µM. In terms of the mechanism, these compounds interacted with DNA causing fragmentation and cell cycle arrest at sub G0/G1 stage. Mutagenesis was higher for NBQ 38 and moderate for NBQ 95 Mitochon-drial permeabilization was observed with NBQ 38 and slightly for NBQ 95. Both compounds caused activation of Caspases 3 and 7 suggesting an apoptotic cell death pathway through an intrinsic mechanism. This study reports evidence of the toxicity of these novel compounds with overlapping structural and mechanistic similarities to ellipticine, a known anti-tumor compound.

Optimization of high throughput virtual screening by combining shape-matching and docking methods.

Receptor flexibility is a critical issue in structure-based virtual screening methods. Although a multiple-receptor conformation docking is an efficient way to account for receptor flexibility, it is still too slow for large molecular libraries. It was reported that a fast ligand-centric, shape-based virtual screening was more consistent for hit enrichment than a typical single-receptor conformation docking. Thus, we designed a "distributed docking" method that improves virtual high throughput screening by combining a shape-matching method with a multiple-receptor conformation docking. Database compounds are classified in advance based on shape similarities to one of the crystal ligands complexed with the target protein. This classification enables us to pick the appropriate receptor conformation for a single-receptor conformation docking of a given compound, thereby avoiding time-consuming multiple docking. In particular, this approach utilizes cross-docking scores of known ligands to all available receptor structures in order to optimize the algorithm. The present virtual screening method was tested for reidentification of known PPARgamma and p38 MAP kinase active compounds. We demonstrate that this method improves the enrichment while maintaining the computation speed of a typical single-receptor conformation docking.

Surrogate docking: structure-based virtual screening at high throughput speed.

Structure-based screening using fully flexible docking is still too slow for large molecular libraries. High quality docking of a million molecule library can take days even on a cluster with hundreds of CPUs. This performance issue prohibits the use of fully flexible docking in the design of large combinatorial libraries. We have developed a fast structure-based screening method, which utilizes docking of a limited number of compounds to build a 2D QSAR model used to rapidly score the rest of the database. We compare here a model based on radial basis functions and a Bayesian categorization model. The number of compounds that need to be actually docked depends on the number of docking hits found. In our case studies reasonable quality models are built after docking of the number of molecules containing approximately 50 docking hits. The rest of the library is screened by the QSAR model. Optionally a fraction of the QSAR-prioritized library can be docked in order to find the true docking hits. The quality of the model only depends on the training set size - not on the size of the library to be screened. Therefore, for larger libraries the method yields higher gain in speed no change in performance. Prioritizing a large library with these models provides a significant enrichment with docking hits: it attains the values of approximately 13 and approximately 35 at the beginning of the score-sorted libraries in our two case studies: screening of the NCI collection and a combinatorial libraries on CDK2 kinase structure. With such enrichments, only a fraction of the database must actually be docked to find many of the true hits. The throughput of the method allows its use in screening of large compound collections and in the design of large combinatorial libraries. The strategy proposed has an important effect on efficiency but does not affect retrieval of actives, the latter being determined by the quality of the docking method itself.

Computational identification of proteins for selectivity assays.

At the stage of optimization of a chemical series the compounds are normally assayed for binding or inhibition on the target protein as well as on several proteins from a selectivity panel. These proteins are normally identified on the basis of sequence homology to the target protein. Experimental selectivity data are also taken into account if available. Cases when a nonhomologous protein has a significant affinity to the compound series are going to be missed if the selectivity panel is identified by homology. Experimental data is usually either unavailable or limited to a small fraction of proteins that should be considered. We have developed a computational method of identification of selectivity panel proteins. It is based on the evaluation of binding site similarity to the target protein using docking scores of target-selected molecular probes. These probes are obtained by docking a large library of drug-like compounds to the target protein followed by selecting a diverse subset from the best virtual binders. Docking scores of these probes to other proteins measure binding site similarity to the target. Because the method does not require prior knowledge of either affinities or structures of inhibitors for the target, it can be applied to any protein with known 3D structure. Validation of the method includes rediscovery of nonhomologous proteins that bind common ligands: estradiol, tamoxifen, and riboflavin. Given 3D structures, the method can effectively discriminate proteins with similar binding sites from random proteins independent of sequence homology.

Inactivation of TNF signaling by rationally designed dominant-negative TNF variants.

Tumor necrosis factor (TNF) is a key regulator of inflammatory responses and has been implicated in many pathological conditions. We used structure-based design to engineer variant TNF proteins that rapidly form heterotrimers with native TNF to give complexes that neither bind to nor stimulate signaling through TNF receptors. Thus, TNF is inactivated by sequestration. Dominant-negative TNFs represent a possible approach to anti-inflammatory biotherapeutics, and experiments in animal models show that the strategy can attenuate TNF-mediated pathology. Similar rational design could be used to engineer inhibitors of additional TNF superfamily cytokines as well as other multimeric ligands.

Computational stabilization of human growth hormone.

Recombinant human growth hormone (hGH) is used worldwide for the treatment of pediatric hypopituitary dwarfism and in children suffering from low levels of hGH. It has limited stability in solution, and because of poor oral absorption, is administered by injection, typically several times a week. Development has therefore focused on more stable or sustained-release formulations and alternatives to injectable delivery that would increase bioavailability and make it easier for patients to use. We redesigned hGH computationally to improve its thermostability. A more stable variant of hGH could have improved pharmacokinetics or enhanced shelf-life, or be more amenable to use in alternate delivery systems and formulations. The computational design was performed using a previously developed combinatorial optimization algorithm based on the dead-end elimination theorem. The algorithm uses an empirical free energy function for scoring designed sequences. This function was augmented with a term that accounts for the loss of backbone and side-chain conformational entropy. The weighting factors for this term, the electrostatic interaction term, and the polar hydrogen burial term were optimized by minimizing the number of mutations designed by the algorithm relative to wild-type. Forty-five residues in the core of the protein were selected for optimization with the modified potential function. The proteins designed using the developed scoring function contained six to 10 mutations, showed enhancement in the melting temperature of up to 16 degrees C, and were biologically active in cell proliferation studies. These results show the utility of our free energy function in automated protein design.