Search

In recent years, reinforcement learning (RL) has emerged as a valuable tool in drug design, offering the potential to propose and optimize molecules with desired properties. However, striking a balance between capabilities, flexibility, reliability, and efficiency remains challenging due to the complexity of advanced RL algorithms and the significant reliance on specialized code. In this work, we introduce ACEGEN, a comprehensive and streamlined toolkit tailored for generative drug design, built using TorchRL, a modern RL library that offers thoroughly tested reusable components. We validate ACEGEN by benchmarking against other published generative modeling algorithms and show comparable or improved performance. We also show examples of ACEGEN applied in multiple drug discovery case studies. ACEGEN is accessible at https://github.com/acellera/acegen-openand available for use under the MIT license.

The management of chronic myeloid leukemia (CML) diagnosed during pregnancy is a rare and challenging situation. We report the treatment and outcome of 87 cases diagnosed in chronic phase from 2001–2022 derived from the largest international observational registry, supported by the European LeukemiaNet (ELN), of 400 pregnancies in 299 CML women. Normal childbirth occurred in 76% without an increased rate of birth abnormalities or life-threatening events, including in patients untreated or treated with interferon-α and/or imatinib in 2nd–3rd trimester. The low birth weight rate of 12% was comparable to that seen in the normal population. Elective and spontaneous abortions occurred in 21% and 3%, respectively. The complete hematologic response rate before labor was 95% with imatinib and 47% with interferon only. No disease progression during pregnancy was observed, 28% of the patients switched their therapy at varying times after delivery. Treatment options balance the efficacy and safety for mother and infant: interferon-α can commence in the 1st trimester and continued throughout in cases of good disease control and tolerability. Because of limited placental crossing, selected tyrosine kinase inhibitors (imatinib and nilotinib) seem to be safe and effective options in 2nd and 3rd trimester while hydroxycarbamide offers few benefits.

This letter gives results on improving protein–ligand binding affinity predictions based on molecular dynamics simulations using machine learning potentials with a hybrid neural network potential and molecular mechanics methodology (NNP/MM). We compute relative binding free energies with the Alchemical Transfer Method and validate its performance against established benchmarks and find significant enhancements compared with conventional MM force fields like GAFF2.

Background: Since primary membranous nephropathy is a heterogeneous disease with variable outcomes and multiple possible therapeutic approaches, all 13 Nephrology Units of the Italian region Emilia Romagna decided to analyze their experience in the management of this challenging glomerular disease. Methods: We retrospectively studied 205 consecutive adult patients affected by biopsy-proven primary membranous nephropathy, recruited from January 2010 through December 2017. The primary outcome was patient and renal survival. The secondary outcome was the rate of complete remission and partial remission of proteinuria. Relapse incidence, treatment patterns and adverse events were also assessed. Results: Median (IQR) follow-up was 36 (24–60) months. Overall patient and renal survival were 87.4% after 5 years. At the end of follow-up, 83 patients (40%) had complete remission and 72 patients (35%) had partial remission. Among responders, less than a quarter (23%) relapsed. Most patients (83%) underwent immunosuppressive therapy within 6 months of biopsy. A cyclic regimen of corticosteroid and cytotoxic agents was the most commonly used treatment schedule (63%), followed by rituximab (28%). Multivariable analysis showed that the cyclic regimen significantly correlates with complete remission (odds ratio 0.26; 95% CI 0.08–0.79) when compared to rituximab (p< 0.05). Conclusions: In our large study, both short- and long-term outcomes were positive and consistent with those published in the literature. Our data suggest that the use of immunosuppressive therapy within the first 6 months after biopsy appears to be a winning strategy, and that the cyclic regimen also warrants a prominent role in primary membranous nephropathy treatment, since definitive proof of rituximab superiority is lacking. Graphic Abstract:

Machine learning plays an important and growing role in molecular simulation. The newest version of the OpenMM molecular dynamics toolkit introduces new features to support the use of machine learning potentials. Arbitrary PyTorch models can be added to a simulation and used to compute forces and energy. A higher-level interface allows users to easily model their molecules of interest with general purpose, pretrained potential functions. A collection of optimized CUDA kernels and custom PyTorch operations greatly improves the speed of simulations. We demonstrate these features in simulations of cyclin-dependent kinase 8 (CDK8) and the green fluorescent protein chromophore in water. Taken together, these features make it practical to use machine learning to improve the accuracy of simulations with only a modest increase in cost.

Developing accurate and efficient coarse-grained representations of proteins is crucial for understanding their folding, function, and interactions over extended time scales. Our methodology involves simulating proteins with molecular dynamics and utilizing the resulting trajectories to train a neural network potential through differentiable trajectory reweighting. Remarkably, this method requires only the native conformation of proteins, eliminating the need for labeled data derived from extensive simulations or memory-intensive end-to-end differentiable simulations. Once trained, the model can be employed to run parallel molecular dynamics simulations and sample folding events for proteins both within and beyond the training distribution, showcasing its extrapolation capabilities. By applying Markov state models, native-like conformations of the simulated proteins can be predicted from the coarse-grained simulations. Owing to its theoretical transferability and ability to use solely experimental static structures as training data, we anticipate that this approach will prove advantageous for developing new protein force fields and further advancing the study of protein dynamics, folding, and interactions.

Many older patients with acute myeloid leukaemia die or cannot undergo allogeneic haematopoietic stem-cell transplantation (HSCT) due to toxicity caused by intensive chemotherapy. We hypothesised that replacing intensive chemotherapy with decitabine monotherapy could improve outcomes.

Machine learning potentials have emerged as a means to enhance the accuracy of biomolecular simulations. However, their application is constrained by the significant computational cost arising from the vast number of parameters compared with traditional molecular mechanics. To tackle this issue, we introduce an optimized implementation of the hybrid method (NNP/MM), which combines a neural network potential (NNP) and molecular mechanics (MM). This approach models a portion of the system, such as a small molecule, using NNP while employing MM for the remaining system to boost efficiency. By conducting molecular dynamics (MD) simulations on various protein–ligand complexes and metadynamics (MTD) simulations on a ligand, we showcase the capabilities of our implementation of NNP/MM. It has enabled us to increase the simulation speed by ∼5 times and achieve a combined sampling of 1 μs for each complex, marking the longest simulations ever reported for this class of simulations.

•An improved pegaspargase-modified MRD and risk-oriented regimen was prospectively tested in adults aged 18-65 years with Ph–ALL/LL.•Three-year survival rates were >60% with HCT or chemotherapy, further improved by MRD negativity and age ≤40 years.

Intrinsically disordered proteins participate in many biological processes by folding upon binding to other proteins. However, coupled folding and binding processes are not well understood from an atomistic point of view. One of the main questions is whether folding occurs prior to or after binding. Here we use a novel, unbiased, high-throughput adaptive sampling approach to reconstruct the binding and folding between the disordered transactivation domain of c-Myb and the KIX domain of the CREB-binding protein. The reconstructed long-term dynamical process highlights the binding of a short stretch of amino acids on c-Myb as a folded α-helix. Leucine residues, especially Leu298-Leu302, establish initial native contacts that prime the binding and folding of the rest of the peptide, with a mixture of conformational selection on the N-terminal region with an induced fit of the C-terminal.

Elotuzumab plus lenalidomide and dexamethasone has shown improved progression-free and overall survival versus lenalidomide and dexamethasone in patients with relapsed or refractory multiple myeloma. We aimed to assess these regimens in patients with newly diagnosed multiple myeloma who are ineligible for haematopoietic stem-cell transplantation (HSCT).

The 2017 version of the European LeukemiaNet (ELN) recommendations, by integrating cytogenetics and mutational status of specific genes, divides patients with acute myeloid leukemia into 3 prognostically distinct risk categories: favorable (ELN2017-FR), intermediate (ELN2017-IR), and adverse (ELN2017-AR). We performed a post hoc analysis of the GIMEMA (Gruppo Italiano Malattie EMatologiche dell’Adulto) AML1310 trial to investigate the applicability of the ELN2017 risk stratification to our study population. In this trial, after induction and consolidation, patients in complete remission were to receive an autologous stem cell transplant (auto-SCT) if categorized as favorable risk or an allogeneic stem cell transplant (allo-SCT) if adverse risk. Intermediate-risk patients were to receive auto-SCT or allo-SCT based on the postconsolidation levels of measurable residual disease as measured by using flow cytometry. Risk categorization was originally conducted according to the 2009 National Comprehensive Cancer Network recommendations. Among 500 patients, 445 (89%) were reclassified according to the ELN2017 criteria: ELN2017-FR, 186 (41.8%) of 455; ELN2017-IR, 179 (40.2%) of 445; and ELN2017-AR, 80 (18%) of 455. In 55 patients (11%), ELN2017 was not applicable. Two-year overall survival (OS) was 68.8%, 51.3%, 45.8%, and 42.8% for the ELN2017-FR, ELN2017-IR, ELN2017-not classifiable, and ELN2017-AR groups, respectively (P < .001). When comparing the 2 different transplant strategies in each ELN2017 risk category, a significant benefit of auto-SCT over allo-SCT was observed among ELN2017-FR patients (2-year OS of 83.3% vs 66.7%; P = .0421). The 2 transplant procedures performed almost equally in the ELN2017-IR group (2-year OS of 73.9% vs 70.8%; P = .5552). This post hoc analysis of the GIMEMA AML1310 trial confirms that the ELN2017 classification is able to accurately discriminate patients with different outcomes and who may benefit from different transplant strategies. This trial was registered as EudraCT number 2010-023809-36 and at www.clinicaltrials.gov as #NCT01452646.

Tyrosine kinase inhibitors have improved survival for patients with Philadelphia chromosome–positive (Ph+) acute lymphoblastic leukemia (ALL). However, prognosis for old or unfit patients remains poor. In the INCB84344-201 (formerly GIMEMA LAL 1811) prospective, multicenter, phase 2 trial, we tested the efficacy and safety of ponatinib plus prednisone in newly diagnosed patients with Ph+ ALL ≥60 years, or unfit for intensive chemotherapy and stem cell transplantation. Forty-four patients received oral ponatinib 45 mg/d for 48 weeks (core phase), with prednisone tapered to 60 mg/m2/d from days-14-29. Prophylactic intrathecal chemotherapy was administered monthly. Median age was 66.5 years (range, 26-85). The primary endpoint (complete hematologic response [CHR] at 24 weeks) was reached in 38/44 patients (86.4%); complete molecular response (CMR) in 18/44 patients (40.9%) at 24 weeks. 61.4% of patients completed the core phase. As of 24 April 2020, median event-free survival was 14.31 months (95% CI 9.30-22.31). Median overall survival and duration of CHR were not reached; median duration of CMR was 11.6 months. Most common treatment-emergent adverse events (TEAEs) were rash (36.4%), asthenia (22.7%), alanine transaminase increase (15.9%), erythema (15.9%), and γ-glutamyltransferase increase (15.9%). Cardiac and vascular TEAEs occurred in 29.5% (grade ≥3, 18.2%) and 27.3% (grade ≥3, 15.9%), respectively. Dose reductions, interruptions, and discontinuations due to TEAEs occurred in 43.2%, 43.2%, and 27.3% of patients, respectively; 5 patients had fatal TEAEs. Ponatinib and prednisone showed efficacy in unfit patients with Ph+ ALL; however, a lower ponatinib dose may be more appropriate in this population. This trial was registered at www.clinicaltrials.gov as #NCT01641107.

Tyrosine kinase inhibitors have improved survival for patients with Philadelphia chromosome–positive (Ph+) acute lymphoblastic leukemia (ALL). However, prognosis for old or unfit patients remains poor. In the INCB84344-201 (formerly GIMEMA LAL 1811) prospective, multicenter, phase 2 trial, we tested the efficacy and safety of ponatinib plus prednisone in newly diagnosed patients with Ph+ALL ≥60 years, or unfit for intensive chemotherapy and stem cell transplantation. Forty-four patients received oral ponatinib 45 mg/d for 48 weeks (core phase), with prednisone tapered to 60 mg/m2/d from days-14-29. Prophylactic intrathecal chemotherapy was administered monthly. Median age was 66.5 years (range, 26-85). The primary endpoint (complete hematologic response [CHR] at 24 weeks) was reached in 38/44 patients (86.4%); complete molecular response (CMR) in 18/44 patients (40.9%) at 24 weeks. 61.4% of patients completed the core phase. As of 24 April 2020, median event-free survival was 14.31 months (95% CI 9.30-22.31). Median overall survival and duration of CHR were not reached; median duration of CMR was 11.6 months. Most common treatment-emergent adverse events (TEAEs) were rash (36.4%), asthenia (22.7%), alanine transaminase increase (15.9%), erythema (15.9%), and γ-glutamyltransferase increase (15.9%). Cardiac and vascular TEAEs occurred in 29.5% (grade ≥3, 18.2%) and 27.3% (grade ≥3, 15.9%), respectively. Dose reductions, interruptions, and discontinuations due to TEAEs occurred in 43.2%, 43.2%, and 27.3% of patients, respectively; 5 patients had fatal TEAEs. Ponatinib and prednisone showed efficacy in unfit patients with Ph+ALL; however, a lower ponatinib dose may be more appropriate in this population. This trial was registered at www.clinicaltrials.govas #NCT01641107.

Deep learning has been successfully applied to structure-based protein–ligand affinity prediction, yet the black box nature of these models raises some questions. In a previous study, we presented KDEEP, a convolutional neural network that predicted the binding affinity of a given protein–ligand complex while reaching state-of-the-art performance. However, it was unclear what this model was learning. In this work, we present a new application to visualize the contribution of each input atom to the prediction made by the convolutional neural network, aiding in the interpretability of such predictions. The results suggest that KDEEPis able to learn meaningful chemistry signals from the data, but it has also exposed the inaccuracies of the current model, serving as a guideline for further optimization of our prediction tools.

Molecular dynamics simulations provide a mechanistic description of molecules by relying on empirical potentials. The quality and transferability of such potentials can be improved leveraging data-driven models derived with machine learning approaches. Here, we present TorchMD, a framework for molecular simulations with mixed classical and machine learning potentials. All force computations including bond, angle, dihedral, Lennard-Jones, and Coulomb interactions are expressed as PyTorch arrays and operations. Moreover, TorchMD enables learning and simulating neural network potentials. We validate it using standard Amber all-atom simulations, learning an ab initio potential, performing an end-to-end training, and finally learning and simulating a coarse-grained model for protein folding. We believe that TorchMD provides a useful tool set to support molecular simulations of machine learning potentials. Code and data are freely available at github.com/torchmd.

The phase 3 GIMEMA-MMY-3006 trial, which compared bortezomib, thalidomide, and dexamethasone (VTD) combination therapy with thalidomide and dexamethasone (TD) as induction therapy before and consolidation therapy after double autologous haematopoietic stem-cell transplantation (HSCT) for newly diagnosed multiple myeloma, showed the superiority of the triplet regimen over the doublet in terms of increased complete response rate and improved progression-free survival. We report the results from the final analysis of the study.

The extreme dynamic behavior of intrinsically disordered proteins hinders the development of drug-like compounds capable of modulating them. There are several examples of small molecules that specifically interact with disordered peptides. However, their mechanisms of action are still not well understood. Here, we use extensive molecular dynamics simulations combined with adaptive sampling algorithms to perform free ligand binding studies in the context of intrinsically disordered proteins. We tested this approach in the system composed by the D2 sub-domain of the disordered protein p27and the small molecule SJ403. The results show several protein–ligand bound states characterized by the establishment of a loosely oriented interaction mediated by a limited number of contacts between the ligand and critical residues of p27. Finally, protein conformations in the bound state are likely to be explored by the isolated protein too, therefore supporting a model where the addition of the small molecule restricts the available conformational space.

G-protein-coupled receptors (GPCRs) are involved in numerous physiological processes and are the most frequent targets of approved drugs. The explosion in the number of new three-dimensional (3D) molecular structures of GPCRs (3D-GPCRome) over the last decade has greatly advanced the mechanistic understanding and drug design opportunities for this protein family. Molecular dynamics (MD) simulations have become a widely established technique for exploring the conformational landscape of proteins at an atomic level. However, the analysis and visualization of MD simulations require efficient storage resources and specialized software. Here we present GPCRmd (http://gpcrmd.org/), an online platform that incorporates web-based visualization capabilities as well as a comprehensive and user-friendly analysis toolbox that allows scientists from different disciplines to visualize, analyze and share GPCR MD data. GPCRmd originates from a community-driven effort to create an open, interactive and standardized database of GPCR MD simulations.

Sampling from the equilibrium distribution has always been a major problem in molecular simulations due to the very high dimensionality of the conformational space. Over several decades, many approaches have been used to overcome the problem. In particular, we focus on unbiased simulation methods such as parallel and adaptive sampling. Here, we recast adaptive sampling schemes on the basis of multi-armed bandits and develop a novel adaptive sampling algorithm under this framework, AdaptiveBandit. We test it on multiple simplified potentials and in a protein folding scenario. We find that this framework performs similarly to or better than previous methods in every type of test potential. Furthermore, it provides a novel framework to develop new sampling algorithms with better asymptotic characteristics.

SkeleDock is a scaffold docking algorithm which uses the structure of a protein–ligand complex as a template to model the binding mode of a chemically similar system. This algorithm was evaluated in the D3R Grand Challenge 4 pose prediction challenge, where it achieved competitive performance. Furthermore, we show that if crystallized fragments of the target ligand are available then SkeleDock can outperform rDock docking software at predicting the binding mode. This Application Note also addresses the capacity of this algorithm to model macrocycles and deal with scaffold hopping. SkeleDock can be accessed at https://playmolecule.org/SkeleDock/.

Cryptic pockets are protein cavities that remain hidden in resolved apo structures and generally require the presence of a co-crystallized ligand to become visible. Finding new cryptic pockets is crucial for structure-based drug discovery to identify new ways of modulating protein activity and thus expand the druggable space. We present here a new method and associated web application leveraging mixed-solvent molecular dynamics (MD) simulations using benzene as a hydrophobic probe to detect cryptic pockets. Our all-atom MD-based workflow was systematically tested on 18 different systems and 5 additional kinases and represents the largest validation study of this kind. CrypticScout identifies benzene probe binding hotspots on a protein surface by mapping probe occupancy, residence time, and the benzene occupancy reweighed by the residence time. The method is presented to the scientific community in a web application available via www.playmolecule.orgusing a distributed computing infrastructure to perform the simulations.

Chemical space is impractically large, and conventional structure-based virtual screening techniques cannot be used to simply search through the entire space to discover effective bioactive molecules. To address this shortcoming, we propose a generative adversarial network to generate, rather than search, diverse three-dimensional ligand shapes complementary to the pocket. Furthermore, we show that the generated molecule shapes can be decoded using a shape-captioning network into a sequence of SMILES enabling directly the structure-based de novo drug design. We evaluate the quality of the method by both structure- (docking) and ligand-based [quantitative structure–activity relationship (QSAR)] virtual screening methods. For both evaluation approaches, we observed enrichment compared to random sampling from initial chemical space of ZINC drug-like compounds.

We designed a trial in which postremission therapy of young patients with de novo acute myeloid leukemia (AML) was decided combining cytogenetics/genetics and postconsolidation levels of minimal residual disease (MRD). After induction and consolidation, favorable-risk patients (FR) were to receive autologous stem cell transplant (AuSCT) and poor-risk patients (PR) allogeneic stem cell transplant (AlloSCT). Intermediate-risk patients (IR) were to receive AuSCT or AlloSCT depending on the postconsolidation levels of MRD. Three hundred sixty-one of 500 patients (72%) achieved a complete remission, 342/361 completed the consolidation phase and were treatment allocated: 165 (48%) to AlloSCT (122 PR, 43 IR MRD-positive) plus 23 rescued after salvage therapy, for a total of 188 candidates; 150 (44%) to AuSCT (115 FR, 35 IR MRD-negative) plus 27 IR patients (8%) with no leukemia-associated phenotype, for a total of 177 candidates. Overall, 110/177 (62%) and 130/188 (71%) AuSCT or AlloSCT candidates received it, respectively. Two-year overall (OS) and disease-free survival (DFS) of the whole series was 56% and 54%, respectively. Two-year OS and DFS were 74% and 61% in the FR category, 42% and 45% in the PR category, 79% and 61% in the IR MRD-negative category, and 70% and 67% in the IR MRD-positive category. In conclusion, AuSCT may still have a role in FR and IR MRD-negative categories. In the IR MRD-positive category, AlloSCT prolongs OS and DFS to equal those of the FR category. Using all the available sources of stem cells, AlloSCT was delivered to 71% of the candidates.This trial was registered at www.clinicaltrials.gov as #NCT01452646 and EudraCT as #2010-023809-36.

We designed a trial in which postremission therapy of young patients with de novo acute myeloid leukemia (AML) was decided combining cytogenetics/genetics and postconsolidation levels of minimal residual disease (MRD). After induction and consolidation, favorable-risk patients (FR) were to receive autologous stem cell transplant (AuSCT) and poor-risk patients (PR) allogeneic stem cell transplant (AlloSCT). Intermediate-risk patients (IR) were to receive AuSCT or AlloSCT depending on the postconsolidation levels of MRD. Three hundred sixty-one of 500 patients (72%) achieved a complete remission, 342/361 completed the consolidation phase and were treatment allocated: 165 (48%) to AlloSCT (122 PR, 43 IR MRD-positive) plus 23 rescued after salvage therapy, for a total of 188 candidates; 150 (44%) to AuSCT (115 FR, 35 IR MRD-negative) plus 27 IR patients (8%) with no leukemia-associated phenotype, for a total of 177 candidates. Overall, 110/177 (62%) and 130/188 (71%) AuSCT or AlloSCT candidates received it, respectively. Two-year overall (OS) and disease-free survival (DFS) of the whole series was 56% and 54%, respectively. Two-year OS and DFS were 74% and 61% in the FR category, 42% and 45% in the PR category, 79% and 61% in the IR MRD-negative category, and 70% and 67% in the IR MRD-positive category. In conclusion, AuSCT may still have a role in FR and IR MRD-negative categories. In the IR MRD-positive category, AlloSCT prolongs OS and DFS to equal those of the FR category. Using all the available sources of stem cells, AlloSCT was delivered to 71% of the candidates.This trial was registered at www.clinicaltrials.govas #NCT01452646 and EudraCT as #2010-023809-36.

Fast and accurate molecular force field (FF) parameterization is still an unsolved problem. Accurate FF are not generally available for all molecules, like novel druglike molecules. While methods based on quantum mechanics (QM) exist to parameterize them with better accuracy, they are computationally expensive and slow, which limits applicability to a small number of molecules. Here, we present an automated FF parameterization method which can utilize either density functional theory (DFT) calculations or approximate QM energies produced by different neural network potentials (NNPs), to obtain improved parameters for molecules. We demonstrate that for the case of torchani-ANI-1x NNP, we can parameterize small molecules in a fraction of time compared with an equivalent parameterization using DFT QM calculations while producing more accurate parameters than FF (GAFF2). We expect our method to be of critical importance in computational structure-based drug discovery (SBDD). The current version is available at PlayMolecule(www.playmolecule.org) and implemented in HTMD, allowing to parameterize molecules with different QM and NNP options.

Atomistic or ab initio molecular dynamics simulations are widely used to predict thermodynamics and kinetics and relate them to molecular structure. A common approach to go beyond the time- and length-scales accessible with such computationally expensive simulations is the definition of coarse-grained molecular models. Existing coarse-graining approaches define an effective interaction potential to match defined properties of high-resolution models or experimental data. In this paper, we reformulate coarse-graining as a supervised machine learning problem. We use statistical learning theory to decompose the coarse-graining error and cross-validation to select and compare the performance of different models. We introduce CGnets, a deep learning approach, that learns coarse-grained free energy functions and can be trained by a force-matching scheme. CGnets maintain all physically relevant invariances and allow one to incorporate prior physics knowledge to avoid sampling of unphysical structures. We show that CGnets can capture all-atom explicit-solvent free energy surfaces with models using only a few coarse-grained beads and no solvent, while classical coarse-graining methods fail to capture crucial features of the free energy surface. Thus, CGnets are able to capture multibody terms that emerge from the dimensionality reduction.

All-atom molecular simulations offer detailed insights into macromolecular phenomena, but their substantial computational cost hinders the exploration of complex biological processes. We introduce Advanced Machine-learning Atomic Representation Omni-force-field (AMARO), a new neural network potential (NNP) that combines an O(3)-equivariant message-passing neural network architecture, TensorNet, with a coarse-graining map that excludes hydrogen atoms. AMARO demonstrates the feasibility of training coarser NNP, without prior energy terms, to run stable protein dynamics with scalability and generalization capabilities.