SAMPL6 challenge results from $$pK_a$$ predictions based on a general Gaussian process model

A variety of fields would benefit from accurate \(pK_a\) predictions, especially drug design due to the effect a change in ionization state can have on a molecule’s physiochemical properties. Participants in the recent SAMPL6 blind challenge were asked to submit predictions for microscopic and macroscopic \(pK_a\)s of 24 drug like small molecules. We recently built a general model for predicting \(pK_a\)s using a Gaussian process regression trained using physical and chemical features of each ionizable group. Our pipeline takes a molecular graph and uses the OpenEye Toolkits to calculate features describing the removal of a proton. These features are fed into a Scikit-learn Gaussian process to predict microscopic \(pK_a\)s which are then used to analytically determine macroscopic \(pK_a\)s. Our Gaussian process is trained on a set of 2700 macroscopic \(pK_a\)s from monoprotic and select diprotic molecules. Here, we share our results for microscopic and macroscopic predictions in the SAMPL6 challenge. Overall, we ranked in the middle of the pack compared to other participants, but our fairly good agreement with experiment is still promising considering the challenge molecules are chemically diverse and often polyprotic while our training set is predominately monoprotic. Of particular importance to us when building this model was to include an uncertainty estimate based on the chemistry of the molecule that would reflect the likely accuracy of our prediction. Our model reports large uncertainties for the molecules that appear to have chemistry outside our domain of applicability, along with good agreement in quantile–quantile plots, indicating it can predict its own accuracy. The challenge highlighted a variety of means to improve our model, including adding more polyprotic molecules to our training set and more carefully considering what functional groups we do or do not identify as ionizable.     

Measuring experimental cyclohexane-water distribution coefficients for the SAMPL5 challenge

Small molecule distribution coefficients between immiscible nonaqueuous and aqueous phases—such as cyclohexane and water—measure the degree to which small molecules prefer one phase over another at a given pH. As distribution coefficients capture both thermodynamic effects (the free energy of transfer between phases) and chemical effects (protonation state and tautomer effects in aqueous solution), they provide an exacting test of the thermodynamic and chemical accuracy of physical models without the long correlation times inherent to the prediction of more complex properties of relevance to drug discovery, such as protein-ligand binding affinities. For the SAMPL5 challenge, we carried out a blind prediction exercise in which participants were tasked with the prediction of distribution coefficients to assess its potential as a new route for the evaluation and systematic improvement of predictive physical models. These measurements are typically performed for octanol-water, but we opted to utilize cyclohexane for the nonpolar phase. Cyclohexane was suggested to avoid issues with the high water content and persistent heterogeneous structure of water-saturated octanol phases, since it has greatly reduced water content and a homogeneous liquid structure. Using a modified shake-flask LC-MS/MS protocol, we collected cyclohexane/water distribution coefficients for a set of 53 druglike compounds at pH 7.4. These measurements were used as the basis for the SAMPL5 Distribution Coefficient Challenge, where 18 research groups predicted these measurements before the experimental values reported here were released. In this work, we describe the experimental protocol we utilized for measurement of cyclohexane-water distribution coefficients, report the measured data, propose a new bootstrap-based data analysis procedure to incorporate multiple sources of experimental error, and provide insights to help guide future iterations of this valuable exercise in predictive modeling.     

Laser-induced fluorescence studies of polycyclic aromatic hydrocarbons (PAH) vapors at high temperatures.

In this work, we present the fluorescence spectra of anthracene and pyrene vapors at different elevated temperatures (from 150 to 650 degrees C) excited with the 337 nm line of a nitrogen laser. We describe the high temperature effects on the resulting spectral properties including spectral intensity, spectral bandwidth and spectral shift. We found that the PAH fluorescence spectral bandwidths become very broad as the temperature increases. The broadening is mainly due to thermal vibrational sequence congestion. We also have found that the fluorescence intensity of pyrene vapor increases with increasing temperature, which results from the increase of the pyrene vapor absorption cross section at 337 nm.     

Interrogating HIV integrase for compounds that bind- a SAMPL challenge

Tremendous gains and novel methods are often developed when people are challenged to do something new or difficult. This process is enhanced when people compete against each other-this can be seen in sport as well as in science and technology (e.g. the space race). The SAMPL challenges, like the CASP challenges, aim to challenge modellers and software developers to develop new ways of looking at molecular interactions so the community as a whole can progress in the accurate prediction of these interactions. In order for this challenge to occur, data must be supplied so the prospective test can be done. We have supplied unpublished data related to a drug discovery program run several years ago on HIV integrase for the SAMPL4 challenge. This paper describes the methods used to obtain these data and the chemistry involved.     

Three-dimensional concentration field measurements in a mixing layer using magnetic resonance imaging

Magnetic resonance imaging (MRI) was used to measure the three-dimensional, time-averaged concentration distribution in a turbulent two-stream mixing layer. Test fluids and MRI scanning parameters were chosen to give good signal linearity, and a calibration/normalization procedure was developed to reduce the concentration measurement uncertainty. Plain deionized water mixing with a solution of 0.8% gadopentetate dimeglumine in deionized water were selected as test fluids. The concentration of the marked water was measured on an array of 220,000 0.69 mm³ voxels covering the entire flow apparatus. Planar laser-induced fluorescence experiments were performed on the flow centerplane to provide validation data. The uncertainty of a single voxel measurement was estimated to be less than 12% with the largest source of uncertainty being turbulent dephasing. Averaging two runs in which the marked water was switched between the two streams reduced the uncertainty to only 4%. The complete magnetic resonance concentration (MRC) procedure including the adjustment of scanning parameters, a background run, two reference/calibration runs, and multiple concentration measurement runs can be completed in 2–3 h. This work establishes MRC as a viable technique for studying the mixing in complex turbulent liquid flows.     

Finite-bandwidth effects on the causal prediction of ultrasonic attenuation of the power-law form

Kramers-Kronig (K-K) relations exist as a consequence of causality, placing nonlocal constraints on the relationship between dispersion and absorption. The finite-bandwidth method of applying these relations is examined where the K-K integrals are restricted to the spectrum of the experimental data. These finite-bandwidth K-K relations are known to work with resonant-type data and here are applied to dispersion data consistent with a power-law attenuation coefficient (exponent from 1 to 2). Bandwidth-restricted forms of the zero and once-subtracted K-K relations are used to determine the attenuation coefficient from phase velocity. Analytically, it is shown that these transforms produce the proper power-law form of the attenuation coefficient as a stand-alone term summed with artifacts that are dependent on the integration limits. Calculations are performed to demonstrate how these finite-bandwidth artifacts affect the K-K predictions under a variety of conditions. The predictions are studied in a local context as a function of subtraction frequency, bandwidth, and power-law exponent. The K-K predictions of the power-law exponent within various decades of the spectrum are also examined. In general, the agreement between finite-bandwidth K-K predictions and exact values grows as the power-law exponent approaches 1 and with increasing bandwidth.     

Processing MALDI Mass Spectra to Improve Mass Spectral Direct Tissue Analysis

Profiling and imaging biological specimens using MALDI mass spectrometry has significant potential to contribute to our understanding and diagnosis of disease. The technique is efficient and high-throughput providing a wealth of data about the biological state of the sample from a very simple and direct experiment. However, in order for these techniques to be put to use for clinical purposes, the approaches used to process and analyze the data must improve. This study examines some of the existing tools to baseline subtract, normalize, align, and remove spectral noise for MALDI data, comparing the advantages of each. A preferred workflow is presented that can be easily implemented for data in ASCII format. The advantages of using such an approach are discussed for both molecular profiling and imaging mass spectrometry.     

EPR and magnetic properties of heteronuclear Mn(n)Mg(6-)(n)(O(2)CNEt(2))(12): impact of structural distortions on Mn(II) in weak ligand fields

The reaction between Mn(6)L(12) and Mg(6)L(12) (L = N,N-diethylcarbamate) results in isolation of heteronuclear complexes Mn(n)Mg(6)(-)(n)L(12). A series was prepared with different doping factors n by varying the Mn/Mg ratio in the crystallization solutions. Single-crystal X-ray diffraction shows that MnMg(5)L(12) is isostructural with Mn(6)L(12) and Mg(6)L(12). Magnetic susceptibility data on the series Mn(n)Mg(6)(-)(n)L(12) (n = 1-6) are consistent with antiferromagnetic Mn.Mn interactions. At low n, the magnetic data demonstrate the formation of magnetically isolated Mn(2+) centers. This was confirmed by measurement of the EPR spectrum at a doping factor n = 0.06 in solution, as a powder, and as single crystals. These show hyperfine interactions consistent with isolated Mn(2+). The EPR spectrum of Mn(0.06)Mg(5.94)L(12) exhibits a dominant signal at g(eff) = 4, and a wide series of less intense signals spanning 200-6000 G in the X-band regime. This unusual behavior in a weak-field Mn(2+) complex is attributed to the substantial distortions from cubic ligand field geometry in this system. The g(eff) = 4 signals are attributed to a C(2)-symmetric hexacoordinate Mn(2+) ion with D > 0.3 cm(-)(1) and E/D = 0.33. The wide series is assigned to an axial C(4)(v) pentacoordinate Mn(2+) site with D = 0.05 cm(-)(1). Comparison of the g(eff) = 4 signals to the g = 4.1 signals exhibited by the tetramanganese complex in photosystem II belies the fact that they almost certainly arise from different spin systems. In addition, the similarity of the spectrum of Mn(n)Mg(6)(-)(n)L(12) to mononuclear Mn(4+) complexes suggests that considerable care must be exercised in the use of EPR as a fingerprint for the manganese oxidation state, particularly in manganese proteins where molecular composition may not be precisely established.     

Anomalous magnetic susceptibilities of the alkaline earth hexammines

The magnetic susceptibility of the metallic compounds Ca(NH₃)₆, Sr(NH₃)₆ and Ba(NH₃)₆ has been measured by the Faraday method in the range 2–200 K. The susceptibility of Ca(NH₃)₆ is strongly dependent on temperature and exhibits a broad minimum near 120 K and a peak near 10 K. In contrast to Ca(NH₃)₆, the susceptibilities of Sr(NH₃)₆ and Ba(NH₃)₆ are diamagnetic and decrease rapidly as the temperature is lowered. At low temperatures Ba(NH₃)₆ has the largest mass susceptibility of any nonsuperconducting metal.     

Lead optimization mapper: automating free energy calculations for lead optimization

Alchemical free energy calculations hold increasing promise as an aid to drug discovery efforts. However, applications of these techniques in discovery projects have been relatively few, partly because of the difficulty of planning and setting up calculations. Here, we introduce lead optimization mapper, LOMAP, an automated algorithm to plan efficient relative free energy calculations between potential ligands within a substantial library of perhaps hundreds of compounds. In this approach, ligands are first grouped by structural similarity primarily based on the size of a (loosely defined) maximal common substructure, and then calculations are planned within and between sets of structurally related compounds. An emphasis is placed on ensuring that relative free energies can be obtained between any pair of compounds without combining the results of too many different relative free energy calculations (to avoid accumulation of error) and by providing some redundancy to allow for the possibility of error and consistency checking and provide some insight into when results can be expected to be unreliable. The algorithm is discussed in detail and a Python implementation, based on both Schrödinger’s and OpenEye’s APIs, has been made available freely under the BSD license.