Virtual Screening in the Cloud: How Big Is Big Enough?

Virtual screening is a standard tool in Computer-Assisted Drug Design (CADD). Early in a project, it is typical to use ligand-based similarity search methods to find suitable hit molecules. However, the number of compounds which can be screened and the time required are usually limited by computational resources. We describe here a high-throughput virtual screening project using 3D similarity (FastROCS) and automated evaluation workflows on Orion, a cloud computing platform. Cloud resources make this approach fully scalable and flexible, allowing the generation and search of billions of virtual molecules, and give access to an explicit 3D virtual chemistry space not available before. We discuss the impact of the size of the search space with respect to finding novel chemical hits and the size of the required hit list, as well as computational and economical aspects of resource scaling.

Leaving us with fond memories, smiles, SMILES and, alas, tears: a tribute to David Weininger, 1952-2016.

David Weininger's career, accomplishments, genius, and friendship are warmly remembered by several of his colleagues, friends, and admirers.

The statistics of virtual screening and lead optimization.

Analytic formulae are used to estimate the error for two virtual screening metrics, enrichment factor and area under the ROC curve. These analytic error estimates are then compared to bootstrapping error estimates, and shown to have excellent agreement with respect to area under the ROC curve and good agreement with respect to enrichment factor. The major advantage of the analytic formulae is that they are trivial to calculate and depend only on the number of actives and inactives and the measured value of the metric, information commonly reported in papers. In contrast to this, the bootstrapping method requires the individual compound scores. Methods for converting the error, which is calculated as a variance, into more familiar error bars are also discussed.

Factors determining penetrance in familial atypical haemolytic uraemic syndrome.

BACKGROUND: Inherited abnormalities of complement are found in ∼60% of patients with atypical haemolytic uraemic syndrome (aHUS). Such abnormalities are not fully penetrant. In this study, we have estimated the penetrance of the disease in three families with a CFH mutation (c.3643C>G; p. Arg1215Gly) in whom a common lineage is probable. 25 individuals have been affected with aHUS with three peaks of incidence-early childhood (n=6), early adulthood (n=11) and late adulthood (n=8). Eighteen individuals who have not developed aHUS carry the mutation. METHODS: We estimated penetrance at the ages of 4, 27, 60 and 70 years as both a binary and a survival trait using MLINK and Mendel. We genotyped susceptibility factors in CFH, CD46 and CFHR1 in affected and unaffected carriers. RESULTS AND CONCLUSIONS: We found that the estimates of penetrance at the age of 4 years ranged from <0.01 to 0.10, at the age of 27 years from 0.16 to 0.29, at the age of 60 years from 0.39 to 0.51 and at the age of 70 years from 0.44 to 0.64. We found that the CFH haplotype on the allele not carrying the CFH mutation had a significant effect on disease penetrance. In this family, we did not find that the CD46 haplotypes had a significant effect on penetrance.

Efficient calculation of SAMPL4 hydration free energies using OMEGA, SZYBKI, QUACPAC, and Zap TK.

Several submissions for the SAMPL4 hydration free energy set were calculated using OpenEye tools, including many that were among the top performing submissions. All of our best submissions used AM1BCC charges and Poisson-Boltzmann solvation. Three submissions used a single conformer for calculating the hydration free energy and all performed very well with mean unsigned errors ranging from 0.94 to 1.08 kcal/mol. These calculations were very fast, only requiring 0.5-2.0 s per molecule. We observed that our two single-conformer methodologies have different types of failure cases and that these differences could be exploited for determining when the methods are likely to have substantial errors.

Conformer generation with OMEGA: learning from the data set and the analysis of failures.

We recently published a high quality validation set for testing conformer generators, consisting of structures from both the PDB and the CSD (Hawkins, P. C. D. et al. J. Chem. Inf. Model. 2010, 50, 572.), and tested the performance of our conformer generator, OMEGA, on these sets. In the present publication, we focus on understanding the suitability of those data sets for validation and identifying and learning from OMEGA's failures. We compare, for the first time we are aware of, the coverage of the applicable property spaces between the validation data sets we used and the parent compound sets to determine if our data sets adequately sample these property spaces. We also introduce the concept of torsion fingerprinting and compare this method of dissimilation to the more traditional graph-centric diversification methods we used in our previous publication. To improve our ability to programmatically identify cases where the crystallographic conformation is not well reproduced computationally, we introduce a new metric to compare conformations, RMSTanimoto. This new metric is used alongside those from our previous publication to efficiently identify reproduction failures. We find RMSTanimoto to be particularly effective in identifying failures for the smallest molecules in our data sets. Analysis of the nature of these failures, particularly those for the CSD, sheds further light on the issue of strain in crystallographic structures. Some of the residual failure cases not resolved by simple changes in OMEGA's defaults present significant challenges to conformer generation engines like OMEGA and are a source of new avenues to further improve their performance, while others illustrate the pitfalls of validating against crystallographic ligand conformations, particularly those from the PDB.

Call for Papers: GRC, CADD, and statistics, and all that.

Molecular modeling and the art of computer-aided drug discovery seldom make much use of statistics, despite being fields that can not calculate important properties with great reliability. The 2013 CADD Gordon conference intends to examine what prevents a more effective use of statistics in routine modeling and to raise consciousness as to what is possible. Practical methods will be discussed, deeper issues in applying standard approaches addressed and research on successes and failures in other disciplines presented by invited experts.

The character of molecular modeling.

In asking what progress might occur in molecular modeling in the next 25 years it is worth asking what progress has been made in the last twenty-five. In doing so it is hard to be optimistic for the future of the field unless a greater commitment is made to basic science.

Application of the Gaussian dielectric boundary in Zap to the prediction of protein pKa values.

The results of two rounds of blind pK(a) predictions for ionizable residues in staphylococcal nuclease using OpenEye's legacy protein pK(a) prediction program based on the Zap Poisson-Boltzmann solver were submitted to the 2009 prediction challenge organized by the Protein pK(a) Cooperative and first round predictions were discussed at the corresponding June 2009 Telluride conference. To better understand these results, 21 additional sets of predictions were performed with the same program, varying the internal dielectric, reference pK(a), partial charge set, and dielectric boundary. The internal dielectric (ε(p)) and dielectric boundary were the two most important factors contributing to the quality of the predictions. Although the lowest overall errors were observed with a molecular dielectric boundary at ε(p) = 8, predictions using a smooth Gaussian dielectric boundary performed almost as well at lower ε(p) values because the Gaussian boundary implicitly accounts for a significant level of solvent penetration. Improved pK(a) predictions with the Gaussian boundary methodology will require better prediction and modeling of structural changes due to changes in ionization state, perhaps without resorting to the more exhaustive sampling of conformational states used by other recent continuum methods.

SAMPL2 and continuum modeling.

An account is given of our contributions to the SAMPL2 challenge for vacuum-water transfer energies. These contributions include different charge sets and radii used with Poisson-Boltzmann continuum theory applied to a single low-energy conformation. A rationale for this approach is given, including a summary of what we have learnt over previous SAMPL events. The results strongly suggest the need for new and repeated experimental measurements, both to clarify what appears to be experimental discrepancies in older measurements and to advance the field in a statistically sound manner.

Analysis of SM8 and Zap TK calculations and their geometric sensitivity.

A prospective study of aqueous solvation energies was done using the SM8 and Zap TK models for a variety of geometries. CM4M charges calculated with M06 and M06-2X were found to yield similar results for the SM8 model. Zap TK calculations were primarily done with AM1BCC charges but limited attempts to use charges derived from DFT showed promise. The OMEGA application quickly generated conformations that performed well with both solvation models, while the use of computationally expensive DFT optimized geometries yielded increased RMSE and MSE. It is shown that increasing levels of gas phase geometry optimization yield increasingly unfavorable solvation energy for single conformer models.

The SAMPL2 blind prediction challenge: introduction and overview.

The interactions between a molecule and the aqueous environment underpin any process that occurs in solution, from simple chemical reactions to protein-ligand binding to protein aggregation. Fundamental measures of the interaction between molecule and aqueous phase, such as the transfer energy between gas phase and water or the energetic difference between two tautomers of a molecule in solution, remain nontrivial to predict accurately using current computational methods. SAMPL2 represents the third annual blind prediction of transfer energies, and the first time tautomer ratios were included in the challenge. Over 60 sets of predictions were submitted, and each participant also attempted to estimate the error in their predictions, a task that proved difficult for most. The results of this blind assessment of the state of the field for transfer energy and tautomer ratio prediction both indicate where the field is performing well and point out flaws in current methods.

Can we share questions? Performance of questions from different question banks in a single medical school.

BACKGROUND: To use progress testing, a large bank of questions is required, particularly when planning to deliver tests over a long period of time. The questions need not only to be of good quality but also balanced in subject coverage across the curriculum to allow appropriate sampling. Hence as well as creating its own questions, an institution could share questions. Both methods allow ownership and structuring of the test appropriate to the educational requirements of the institution. METHOD: Peninsula Medical School (PMS) has developed a mechanism to validate questions written in house. That mechanism can be adapted to utilise questions from an International question bank International Digital Electronic Access Library (IDEAL) and another UK-based question bank Universities Medical Assessment Partnership (UMAP). These questions have been used in our progress tests and analysed for relative performance. RESULTS: Data are presented to show that questions from differing sources can have comparable performance in a progress testing format. CONCLUSION: There are difficulties in transferring questions from one institution to another. These include problems of curricula and cultural differences. Whilst many of these difficulties exist, our experience suggests that it only requires a relatively small amount of work to adapt questions from external question banks for effective use. The longitudinal aspect of progress testing (albeit summatively) may allow more flexibility in question usage than single high stakes exams.

Convergent validity and reliability of a novel repeated agility protocol in junior rugby league players.

Background: : Rugby league involves repeated, complex, and high intensity change-of-direction (COD) movements with no existing test protocols that specifically assesses these multiple physical fitness components simultaneously. Thus, the current study examined the convergent validity of a repeated Illinois Agility (RIA) protocol with the repeated T-agility protocol, and the repeatability of the RIA protocol in adolescent Rugby League players. Furthermore, aerobic capacity and anaerobic and COD performance were assessed to determine whether these physical qualities were important contributors to the RIA protocol. Methods: Twenty-two junior Rugby League players completed 4 sessions with each separated by 7 days. Initially, physical fitness characteristics at baseline (i.e., Multi-stage Shuttle test, countermovement jump, 30-m sprint, single-effort COD and repeated sprint ability [RSA]) were assessed. The second session involved a familiarisation of RIA and repeated T-agility test (RTT) protocols. During the third and fourth sessions, participants completed the RIA and RTT protocols in a randomised, counterbalanced design to examine the validity and test-retest reliability of these protocols. Results: For convergent validity, significant correlations were identified between RIA and RTT performances (r= >0.80; p<0.05). For contributors to RIA performance, significant correlations were identified between all baseline fitness characteristics and RIA (r = >0.71; p < 0.05). Reliability of the RIA protocol was near perfect with excellent intra-class correlation coefficient (0.87-0.97), good ratio limits of agreement (×/÷ 1.05-1.06) and low coefficient of variations (1.8-2.0%). Conclusions: The current study has demonstrated the RIA to be a simple, valid and reliable field test for RL athletes that can provide coaches with information about their team's ability to sustain high intensity, multi-directional running efforts.

The SAMP1 solvation challenge: further lessons regarding the pitfalls of parametrization.

This paper describes an investigation into the performance of different charges and radii with the Poisson-Boltzmann method for the calculation of vacuum-water transfer energies. The test set consisted of 63 drug-like molecules used in a blind-test challenge. The results are consistent with earlier reports, namely, that more physical charges perform better and that radii parametrization can both improve and also dramatically worsen results, with the latter suggesting a failure to capture all of the basic physics of solvation.

The Effect of a Resistance Training Session on Physiological and Thermoregulatory Measures of Sub-maximal Running Performance in the Heat in Heat-Acclimatized Men.

BACKGROUND: The current study examined the acute effects of a lower body resistance training (RT) session on physiological and thermoregulatory measures during a sub-maximal running protocol in the heat in heat-acclimatized men. Ten resistance-untrained men (age 27.4 ± 4.1 years; height 1.78 ± 0.06 m; body mass 76.8 ± 9.9 kg; peak oxygen uptake 48.2 ± 7.0 mL kg-1 min-1) undertook a high-intensity RT session at six-repetition maximum. Indirect muscle damage markers (i.e., creatine kinase [CK], delayed-onset muscle soreness [DOMS], and countermovement jump [CMJ]) were collected prior to, immediately post and 24 and 48 h after the RT session. The sub-maximal running protocol was performed at 70% of the ventilatory threshold, which was conducted prior to and 24 and 48 h following the RT session to obtain physiological and thermoregulatory measures. RESULTS: The RT session exhibited significant increases in DOMS (p < 0.05; effect size [ES]: 1.41-10.53), whilst reduced CMJ (p < 0.05; ES: - 0.79-1.41) for 48 h post-exercise. There were no differences in CK (p > 0.05), although increased with moderate to large ES (0.71-1.12) for 48 h post-exercise. The physiological cost of running was increased for up to 48 h post-exercise (p < 0.05) with moderate to large ES (0.50-0.84), although no differences were shown in thermoregulatory measures (p > 0.05) with small ES (0.33). CONCLUSION: These results demonstrate that a RT session impairs sub-maximal running performance for several days post-exercise, although thermoregulatory measures are unperturbed despite elevated muscle damage indicators in heat-acclimatized, resistance untrained men. Accordingly, whilst a RT session may not increase susceptibility to heat-related injuries in heat-acclimatized men during sub-maximal running in the heat, endurance sessions should be undertaken with caution for at least 48 h post-exercise following the initial RT session in resistance untrained men.

Predicting small-molecule solvation free energies: an informal blind test for computational chemistry.

Experimental data on the transfer of small molecules between vacuum and water are relatively sparse. This makes it difficult to assess whether computational methods are truly predictive of this important quantity or merely good at explaining what has been seen. To explore this, a prospective test was performed of two different methods for estimating solvation free energies: an implicit solvent approach based on the Poisson-Boltzmann equation and an explicit solvent approach using alchemical free energy calculations. For a set of 17 small molecules, root mean square errors from experiment were between 1.3 and 2.6 kcal/mol, with the explicit solvent free energy approach yielding somewhat greater accuracy but at greater computational expense. Insights from outliers and suggestions for future prospective challenges of this kind are presented.

Comparison of shape-matching and docking as virtual screening tools.

Ligand docking is a widely used approach in virtual screening. In recent years a large number of publications have appeared in which docking tools are compared and evaluated for their effectiveness in virtual screening against a wide variety of protein targets. These studies have shown that the effectiveness of docking in virtual screening is highly variable due to a large number of possible confounding factors. Another class of method that has shown promise in virtual screening is the shape-based, ligand-centric approach. Several direct comparisons of docking with the shape-based tool ROCS have been conducted using data sets from some of these recent docking publications. The results show that a shape-based, ligand-centric approach is more consistent than, and often superior to, the protein-centric approach taken by docking.

Sirolimus-induced remission of posttransplantation lymphoproliferative disorder.

Posttransplantation lymphoproliferative disorder (PTLD) is one of the most serious complications of solid-organ transplantation. It potentially is treatable in most cases, but current methods involve withdrawal or reduction of immunosuppression and the consequent risk for graft rejection. Sirolimus was shown in vivo and in vitro to limit proliferation of a number of malignant cell lines, including those of PTLD-derived cells. We present a case of disseminated PTLD in a patient with a renal transplant that resolved completely with conversion of immunosuppression to sirolimus. Graft function was maintained and improved with treatment. This offers a novel means of treating these patients and minimizing transplant loss.