Philip Prathipati

Comparison of MLR, PLS and GA-MLR in QSAR analysis*

The use of the internet has evolved in quantitative structure–activity relationship (QSAR) over the past decade with the development of web based activities like the availability of numerous public domain software tools for descriptor calculation and chemometric toolboxes. The importance of chemometrics in QSAR has accelerated in recent years for processing the enormous amount of information in form of predictive mathematical models for large datasets of molecules. With the availability of huge numbers of physicochemical and structural parameters, variable selection became crucial in deriving interpretable and predictive QSAR models. Among several approaches to address this problem, the principle component regression (PCR) and partial least squares (PLS) analyses provide highly predictive QSAR models but being more abstract, they are difficult to understand and interpret. Genetic algorithm (GA) is a stochastic method well suited to the problem of variable selection and to solve optimization problems. Consequently the hybrid approach (GA-MLR) combining GA with multiple linear regression (MLR) may be useful in derivation of highly predictive and interpretable QSAR models. In view of the above, a comparative study of stepwise-MLR, PLS and GA-MLR in deriving QSAR models for datasets of α1-adrenoreceptor antagonists and β3-adrenoreceptor agonists has been carried out using the public domain software Dragon for computing descriptors and free Matlab codes for data modeling.

Development of 3D-QSAR models for 5-Lipoxygenase antagonists: chalcones☆

5-Lipoxygenase inhibitors are of current interest for asthma therapy and inflammatory diseases. In order to identify the essential structural and physicochemical requirements in terms of common biophoric sites (pharmacophore) and secondary sites for binding and interacting with 5-lipoxygenase, a series of 51 compounds of chalcones has been used for the development of 3D-QSAR models on APEX-3D expert system. Among several models, the two models have been identified with the statistical criteria R(2)>0.75, Chance <0.001 and Match >0.7. Both the models (nos 1 and 2) with three biophoric sites and four secondary sites, showed very good correlation (r>0.9) between the observed and calculated or predicted activities.

Characterization of β3-adrenergic receptor: determination of pharmacophore and 3D QSAR model for β3 adrenergic receptor agonism

SummaryThe β3-adrenoreceptor (β3-AR) has been shown to mediate various pharmacological and physiological effects such as lipolysis, thermogenesis, and intestinal smooth muscle relaxation. It also plays an important role in glucose homeostasis and energy balance. Molecular modeling studies were undertaken to develop predictive pharmacophoric hypothesis and 3D-QSAR model, which may explain variations in β3-AR agonistic activity in terms of chemical features and physicochemical properties. The two softwares, CATALYST for pharmacophoric alignment and APEX-3D for 3D-QSAR modeling were used to establish the structure activity relationships for β3-AR agonistic activity. Among the several statistically significant models, the selection of the best pharmacophore and 3D-QSAR model was based on its ability to estimate the activity of external test sets of similar and different structural types along with the reasonable consistency of the model with the limited information of the active site of β3-AR. The final 3D-QSAR model was derived using the pharmacophoric alignments from the hypothesis which consisted of four chemical features: basic or positive ionizable feature on the nitrogen of the aryloxypropylamino group, two ring aromatic features corresponding to the phenyl ring of the phenoxide and the benzenesulphonamido groups and a hydrogen-bond donor (HBD) in the vicinity of the nitrogen atom of the benzenesulphonamido group with the most active molecule mapping in an energetically favorable extended conformation. This hypothesis was in agreement with the site directed mutagenesis studies on human β3-AR and correlated well the observed and estimated activity both in, training and both the external test sets. It also mapped reasonably well to six β3-AR agonists of different structural classes under clinical development and thus this hypothesis may have a universal applicability in providing a powerful template for virtual screening and also for designing new chemical entities (NCEs) as β3-AR agonists.

Computer-Aided Drug Design: Integration of Structure-Based and Ligand-Based Approaches in Drug Design

In silico high throughput screens provide an efficient (time and money) and effective (with comparable or better accuracy) alternatives in comparison to their experimental counterparts, and hence is of enormous interest to drug discovery research. However the assessment of a variety of virtual screening techniques ranging from simple fingerprint based similarity searching to the sophisticated docking algorithms reveals the inverse proportionality of the speed and accuracy of these algorithms, thus presenting a significant challenge, in enabling the use of computational tools to drug research. Some of the advantages and disadvantages of the structure-based (direct) and ligand-based (indirect) drug design techniques are typically discussed in terms of their requirements vis-a-vis the accuracy and time required for the analysis. The various integration strategies conceptualized to circumvent the above problems in the recent years are summarized with their merits and demerits.

Evaluation of binary QSAR models derived from LUDI and MOE scoring functions for structure based virtual screening.

In todays world of high-throughput in silico screening, the development of virtual screening methodologies to prioritize small molecules as new chemical entities (NCEs) for synthesis is of current interest. Among several approaches to virtual screening, structure-based virtual screening has been considered the most effective. However the problems associated with the ranking of potential solutions in terms of scoring functions remains one of the major bottlenecks in structure-based virtual screening technology. It has been suggested that scoring functions may be used as filters for distinguishing binders from nonbinders instead of accurately predicting their binding free energies. Subsequently, several improvements have been made in this area, which include the use of multiple rather than single scoring functions and application of either consensus or multivariate statistical methods or both to improve the discrimination between binders and nonbinders. In view of it, the discriminative ability (distinguishing binders from nonbinders) of binary QSAR models derived using LUDI and MOE scoring functions has been compared with the models derived by Jacobbsson et al. on five data sets viz. estrogen receptor alphamimics (ERalpha_mimics), estrogen receptor alphatoxins (ERalpha_toxins), matrix metalloprotease 3 inhibitors (MMP-3), factor Xa inhibitors (fXa), and acetylcholine esterase inhibitors (AChE). The overall analyses reveal that binary QSAR is comparable to the PLS discriminant analysis, rule-based, and Bayesian classification methods used by Jacobsson et al. Further the scoring functions implemented in LUDI and MOE can score a wide range of protein-ligand interactions and are comparable to the scoring functions implemented in ICM and Cscore. Thus the binary QSAR models derived using LUDI and MOE scoring functions may be useful as a preliminary screening layer in a multilayered virtual screening paradigm.

CoMFA and docking studies on glycogen phosphorylase a inhibitors as antidiabetic agents.

Glycogen phosphorylase (GP(a)) is a specific target for the design of inhibitors and may prevent glycogenolysis under high glucose conditions in type II diabetes. The carboxamides first reported by Hoover D. J. et al. (J. Med. Chem. 1998, 41, 2934-2938) are one of the major classes of GP(a) inhibitors other than glucose derivatives. The recent, X-ray crystallographic analyses (Oikonomakos et al. Biochim. Biophys. Acta 2003, 1647, 325-332) have revealed a distinct mechanism of action for these inhibitors, which bind at a new allosteric site away from the inhibitory and catalytic sites. To elucidate the essential structural and physicochemical requirements responsible for binding to the GP(a) enzyme and to develop predictive models, CoMFA and docking studies have been carried out on a series of indole-2-carboxamide derivates. The CoMFA model developed using pharmacophoric alignments and hydrogen-bonding fields demonstrated high predictive ability against the training (r2 = 0.98, q2 = 0.68) and the test set (r2pred = 0.85). Further the superimposition of PLS coefficient contour maps from CoMFA with the GP(a) active site (PDB: 1lwo) has shown a high level of compatibility.

2D-QSAR in hydroxamic acid derivatives as peptide deformylase inhibitors and antibacterial agents.

Peptide deformylase catalyzes the removal of N-formyl group from the N-formylmethionine of ribosome synthesized polypeptide in eubacteria. Quantitative structure-activity relationship (QSAR) studies have been carried out in a series of beta-sulfonyl and beta-sulfinyl hydroxamic acid derivatives for their PDF enzyme inhibitory and antibacterial activities against Escherichia coli DC2 and Moraxella catarrhalis RA21 which demonstrate that the PDF inhibitory activity in cell free and whole cell system increases with increase in molar refractivity and hydrophobicity. The comparison of the QSARs between the cell free and whole cell system indicate that the active binding sites in PDF isolated from E. coli and in M. catarrhalis RA21 are similar and the whole cell antibacterial activity is mainly due to the inhibition of PDF. Apart from this the QSARs on some matrixmetelloproteins (COL-1, COL-3, MAT and HME) and natural endopeptidase (NEP) indicate the possibilities of introducing selectivity in these hydroxamic acid derivatives for their PDF inhibitory activity.

Network analysis and in silico prediction of protein–protein interactions with applications in drug discovery

Protein-protein interactions (PPIs) are vital to maintaining cellular homeostasis. Several PPI dysregulations have been implicated in the etiology of various diseases and hence PPIs have emerged as promising targets for drug discovery. Surface residues and hotspot residues at the interface of PPIs form the core regions, which play a key role in modulating cellular processes such as signal transduction and are used as starting points for drug design. In this review, we briefly discuss how PPI networks (PPINs) inferred from experimentally characterized PPI data have been utilized for knowledge discovery and how in silico approaches to PPI characterization can contribute to PPIN-based biological research. Next, we describe the principles of in silico PPI prediction and survey the existing PPI and PPI site prediction servers that are useful for drug discovery. Finally, we discuss the potential of in silico PPI prediction in drug discovery.

Integration of Ligand and Structure Based Approaches for CSAR-2014.

The prediction of binding poses and affinities is an area of active interest in computer-aided drug design (CADD). Given the documented limitations with either ligand or structure based approaches, we employed an integrated approach and developed a rapid protocol for binding mode and affinity predictions. This workflow was applied to the three protein targets of Community Structure-Activity Resource-2014 (CSAR-2014) exercise: Factor Xa (FXa), Spleen Tyrosine Kinase (SYK), and tRNA (guanine-N(1))-methyltransferase (TrmD). Our docking and scoring workflow incorporates compound clustering and ligand and protein structure based pharmacophore modeling, followed by local docking, minimization, and scoring. While the former part of the protocol ensures high-quality ligand alignments and mapping, the subsequent minimization and scoring provides the predicted binding modes and affinities. We made blind predictions of docking pose for 1, 5, and 14 ligands docked into 1, 2, and 12 crystal structures of FXa, SYK, and TrmD, respectively. The resulting 174 poses were compared with cocrystallized structures (1, 5, and 14 complexes) made available at the end of CSAR. Our predicted poses were related to the experimentally determined structures with a mean root-mean-square deviation value of 3.4 Å. Further, we were able to classify high and low affinity ligands with the area under the curve values of 0.47, 0.60, and 0.69 for FXa, SYK, and TrmD, respectively, indicating the validity of our approach in at least two of the three systems. Detailed critical analysis of the results and CSAR methodology ranking procedures suggested that a straightforward application of our workflow has limitations, as some of the performance measures do not reflect the actual utility of pose and affinity predictions in the biological context of individual systems.

Collection and preparation of molecular databases for virtual screening

Drug discovery and development research is undergoing a paradigm shift from a linear and sequential nature of the various steps involved in the drug discovery process of the past to the more parallel approach of the present, due to a lack of sufficient correlation between activities estimated by in vitro and in vivo assays. This is attributed to the non-drug-likeness of the lead molecules, which has often been detected at advanced drug development stages. Thus a striking aspect of this paradigm shift has been early/parallel in silico prioritization of drug-like molecular databases (also database pre-processing), in addition to prioritizing compounds with high affinity and selectivity for a protein target. In view of this, a drug-like database useful for virtual screening has been created by prioritizing molecules from 36 catalog suppliers, using our recently derived binary QSAR based drug-likeness model as a filter. The performance of this model was assessed by a comparative evaluation with respect to commonly used filters implemented by the ZINC database. Since the model was derived considering all the limitations that have plagued the existing rules and models, it performs better than the existing filters and thus the molecules prioritized by this filter represent a better subset of drug-like compounds. The application of this model on exhaustive subsets of 4,972,123 molecules, many of which have passed the ZINC database filters for drug-likeness, led to a further prioritization of 2,920,551 drug-like molecules. This database may have a great potential for in silico virtual screening for discovering molecules, which may survive the later stages of the drug development research.