SIMPD: an algorithm for generating simulated time splits for validating machine learning approaches.

Time-split cross-validation is broadly recognized as the gold standard for validating predictive models intended for use in medicinal chemistry projects. Unfortunately this type of data is not broadly available outside of large pharmaceutical research organizations. Here we introduce the SIMPD (simulated medicinal chemistry project data) algorithm to split public data sets into training and test sets that mimic the differences observed in real-world medicinal chemistry project data sets. SIMPD uses a multi-objective genetic algorithm with objectives derived from an extensive analysis of the differences between early and late compounds in more than 130 lead-optimization projects run within the Novartis Institutes for BioMedical Research. Applying SIMPD to the real-world data sets produced training/test splits which more accurately reflect the differences in properties and machine-learning performance observed for temporal splits than other standard approaches like random or neighbor splits. We applied the SIMPD algorithm to bioactivity data extracted from ChEMBL and created 99 public data sets which can be used for validating machine-learning models intended for use in the setting of a medicinal chemistry project. The SIMPD code and simulated data sets are available under open-source/open-data licenses at github.com/rinikerlab/molecular_time_series.

PREFER: A New Predictive Modeling Framework for Molecular Discovery.

Machine-learning and deep-learning models have been extensively used in cheminformatics to predict molecular properties, to reduce the need for direct measurements, and to accelerate compound prioritization. However, different setups and frameworks and the large number of molecular representations make it difficult to properly evaluate, reproduce, and compare them. Here we present a new PREdictive modeling FramEwoRk for molecular discovery (PREFER), written in Python (version 3.7.7) and based on AutoSklearn (version 0.14.7), that allows comparison between different molecular representations and common machine-learning models. We provide an overview of the design of our framework and show exemplary use cases and results of several representation-model combinations on diverse data sets, both public and in-house. Finally, we discuss the use of PREFER on small data sets. The code of the framework is freely available on GitHub.

Therapy targeting antigen-specific T cells by a peptide-based tolerizing vaccine against autoimmune arthritis.

A longstanding goal has been to find an antigen-specific preventive therapy, i.e., a vaccine, for autoimmune diseases. It has been difficult to find safe ways to steer the targeting of natural regulatory antigen. Here, we show that the administration of exogenous mouse major histocompatibility complex class II protein bounding a unique galactosylated collagen type II (COL2) peptide (Aq-galCOL2) directly interacts with the antigen-specific TCR through a positively charged tag. This leads to expanding a VISTA-positive nonconventional regulatory T cells, resulting in a potent dominant suppressive effect and protection against arthritis in mice. The therapeutic effect is dominant and tissue specific as the suppression can be transferred with regulatory T cells, which downregulate various autoimmune arthritis models including antibody-induced arthritis. Thus, the tolerogenic approach described here may be a promising dominant antigen-specific therapy for rheumatoid arthritis, and in principle, for autoimmune diseases in general.

Multispecies Machine Learning Predictions of In Vitro Intrinsic Clearance with Uncertainty Quantification Analyses.

In pharmaceutical research, compounds are optimized for metabolic stability to avoid a too fast elimination of the drug. Intrinsic clearance (CLint) measured in liver microsomes or hepatocytes is an important parameter during lead optimization. In this work, machine learning models were developed to relate the compound structure to microsomal metabolic stability and predict CLint for new compounds. A multitask (MT) learning architecture was introduced to model the CLint of six species simultaneously, giving as a result a multispecies machine learning model. MT graph neural network (MT-GNN) regression was identified as the top-performing method, and an ensemble of 10 MT-GNN models was evaluated prospectively. Geometric mean fold errors were consistently smaller than 2-fold. Moreover, high precision values were obtained in the prediction of "high" (>300 µL/min/mg) and "low" (<100 µL/min/mg) CLint compounds. Precision values ranged from 80 to 94% for low CLint predictions and from 75 to 97% for high CLint predictions, depending on the species. Uncertainty on experimental values and model predictions was systematically quantified. Experimental variability (aleatoric uncertainty) of all historical Novartis in vitro clearance experiments was analyzed. Interestingly, MT-GNN models' performance approached assays' experimental variability. Moreover, uncertainty estimation in predictions (epistemic uncertainty) enabled identifying predictions associated with lower and higher error. Taken together, our manuscript combines a multispecies deep learning model and large-scale uncertainty analyses to improve CLint predictions and facilitate early informed decisions for compound prioritization.

Key interactions in the trimolecular complex consisting of the rheumatoid arthritis-associated DRB1*04:01 molecule, the major glycosylated collagen II peptide and the T-cell receptor.

OBJECTIVES: Rheumatoid arthritis (RA) is an autoimmune disease strongly associated with the major histocompatibility complex (MHC) class II allele DRB1*04:01, which encodes a protein that binds self-peptides for presentation to T cells. This study characterises the autoantigen-presenting function of DRB1*04:01 (HLA-DRA*01:01/HLA-DRB1*04:01) at a molecular level for prototypic T-cell determinants, focusing on a post-translationally modified collagen type II (Col2)-derived peptide. METHODS: The crystal structures of DRB1*04:01 molecules in complex with the peptides HSP70289-306, citrullinated CILP982-996 and galactosylated Col2259-273 were determined on cocrystallisation. T cells specific for Col2259-273 were investigated in peripheral blood mononuclear cells from patients with DRB1*04:01-positive RA by cytofluorometric detection of the activation marker CD154 on peptide stimulation and binding of fluorescent DRB1*0401/Col2259-273 tetramer complexes. The cDNAs encoding the T-cell receptor (TCR) α-chains and ß-chains were cloned from single-cell sorted tetramer-positive T cells and transferred via a lentiviral vector into TCR-deficient Jurkat 76 cells. RESULTS: The crystal structures identified peptide binding to DRB1*04:01 and potential side chain exposure to T cells. The main TCR recognition sites in Col2259-273 were lysine residues that can be galactosylated. RA T-cell responses to DRB1*04:01-presented Col2259-273 were dependent on peptide galactosylation at lysine 264. Dynamic molecular modelling of a functionally characterised Col2259-273-specific TCR complexed with DRB1*04:01/Col2259-273 provided evidence for differential allosteric T-cell recognition of glycosylated lysine 264. CONCLUSIONS: The MHC-peptide-TCR interactions elucidated in our study provide new molecular insights into recognition of a post-translationally modified RA T-cell determinant with a known dominant role in arthritogenic and tolerogenic responses in murine Col2-induced arthritis.

Evaluation guidelines for machine learning tools in the chemical sciences.

Machine learning (ML) promises to tackle the grand challenges in chemistry and speed up the generation, improvement and/or ordering of research hypotheses. Despite the overarching applicability of ML workflows, one usually finds diverse evaluation study designs. The current heterogeneity in evaluation techniques and metrics leads to difficulty in (or the impossibility of) comparing and assessing the relevance of new algorithms. Ultimately, this may delay the digitalization of chemistry at scale and confuse method developers, experimentalists, reviewers and journal editors. In this Perspective, we critically discuss a set of method development and evaluation guidelines for different types of ML-based publications, emphasizing supervised learning. We provide a diverse collection of examples from various authors and disciplines in chemistry. While taking into account varying accessibility across research groups, our recommendations focus on reporting completeness and standardizing comparisons between tools. We aim to further contribute to improved ML transparency and credibility by suggesting a checklist of retro-/prospective tests and dissecting their importance. We envisage that the wide adoption and continuous update of best practices will encourage an informed use of ML on real-world problems related to the chemical sciences.

[Presentation of an innovative interdisciplinary and cross-sector therapeutic concept for pain patients as part of a regional selective contract]. / Vorstellung eines innovativen interdisziplinären, sektoren-übergreifenden und gestuften Behandlungskonzepts für Schmerzpatient*innen im Rahmen eines regionalen Selektivvertrages nach §140a SGB V.

BACKGROUND: A graded therapeutic concept for the treatment of chronic pain patients in Germany is only available to a limited extent. Following the onset of coronavirus disease (COVID-19), care for these patients has become even worse. AIM: To develop and establish a cross-sector therapeutic concept for chronic pain patients as part of a selective contract. METHODS: Embedded in existing therapeutic procedures, we define seven clinical pathways (CPs) into which patients are directed, after an interdisciplinary assessment according to refined criteria. ORGANIZATION: In CP I, patients remain in standard therapy. In CP II, patients have the opportunity to participate in an additional inter-profession education program. In CP III, patients get a specialized outpatient treatment. CP IV is a partial inpatient treatment, where multiple inpatient attendance days are replaced by tele-medical treatment, via a rehabilitation app. CP V and VI are inpatient treatments over 8 and 15 days each. If patients need further psychotherapeutic support after an inpatient treatment, they can be treated by clinical psychotherapists for another six months in CP VII. EVALUATION: The evaluation takes place 3, 6 and 12 months after initial assessment and includes the German Pain Questionnaire with different psychometric tests.

Chemical Reactivity Prediction: Current Methods and Different Application Areas.

The ability to predict chemical reactivity of a molecule is highly desirable in drug discovery, both ex vivo (synthetic route planning, formulation, stability) and in vivo: metabolic reactions determine pharmacodynamics, pharmacokinetics and potential toxic effects, and early assessment of liabilities is vital to reduce attrition rates in later stages of development. Quantum mechanics offer a precise description of the interactions between electrons and orbitals in the breaking and forming of new bonds. Modern algorithms and faster computers have allowed the study of more complex systems in a punctual and accurate fashion, and answers for chemical questions around stability and reactivity can now be provided. Through machine learning, predictive models can be built out of descriptors derived from quantum mechanics and cheminformatics, even in the absence of experimental data to train on. In this article, current progress on computational reactivity prediction is reviewed: applications to problems in drug design, such as modelling of metabolism and covalent inhibition, are highlighted and unmet challenges are posed.

Clinical outcomes following intra-articular injection of autologous adipose-derived mesenchymal stem cells for the treatment of osteoarthritis in dogs characterized by weight-bearing asymmetry.

AIMS: This study investigates the effects of intra-articular injection of adipose-derived mesenchymal stem cells (AdMSCs) and platelet-rich plasma (PRP) on lameness, pain, and quality of life in osteoarthritic canine patients. METHODS: With informed owner consent, adipose tissue collected from adult dogs diagnosed with degenerative joint disease was enzymatically digested and cultured to passage 1. A small portion of cells (n = 4) surplus to clinical need were characterized using flow cytometry and tri-lineage differentiation. The impact and degree of osteoarthritis (OA) was assessed using the Liverpool Osteoarthritis in Dogs (LOAD) score, Modified Canine Osteoarthritis Staging Tool (mCOAST), kinetic gait analysis, and diagnostic imaging. Overall, 28 joints (25 dogs) were injected with autologous AdMSCs and PRP. The patients were followed up at two, four, eight, 12, and 24 weeks. Data were analyzed using two related-samples Wilcoxon signed-rank or Mann-Whitney U tests with statistical significance set at p < 0.05. RESULTS: AdMSCs demonstrated stem cell-like characteristics. LOAD scores were significantly lower at week 4 compared with preinjection (p = 0.021). The mCOAST improved significantly after three months (p = 0.001) and six months (p = 0.001). Asymmmetry indices decreased from four weeks post-injection and remained significantly lower at six months (p = 0.025). CONCLUSION: These improvements in quality of life, reduction in pain on examination, and improved symmetry in dogs injected with AdMSCs and PRP support the effectiveness of this combined treatment for symptom modification in canine OA for six months. Cite this article: Bone Joint Res 2021;10(10):650-658.

GHOST: Adjusting the Decision Threshold to Handle Imbalanced Data in Machine Learning.

Machine learning classifiers trained on class imbalanced data are prone to overpredict the majority class. This leads to a larger misclassification rate for the minority class, which in many real-world applications is the class of interest. For binary data, the classification threshold is set by default to 0.5 which, however, is often not ideal for imbalanced data. Adjusting the decision threshold is a good strategy to deal with the class imbalance problem. In this work, we present two different automated procedures for the selection of the optimal decision threshold for imbalanced classification. A major advantage of our procedures is that they do not require retraining of the machine learning models or resampling of the training data. The first approach is specific for random forest (RF), while the second approach, named GHOST, can be potentially applied to any machine learning classifier. We tested these procedures on 138 public drug discovery data sets containing structure-activity data for a variety of pharmaceutical targets. We show that both thresholding methods improve significantly the performance of RF. We tested the use of GHOST with four different classifiers in combination with two molecular descriptors, and we found that most classifiers benefit from threshold optimization. GHOST also outperformed other strategies, including random undersampling and conformal prediction. Finally, we show that our thresholding procedures can be effectively applied to real-world drug discovery projects, where the imbalance and characteristics of the data vary greatly between the training and test sets.

Automated chiral method screening - Evaluation of generated chromatographic data sets to further optimize screening efficiency.

We set up an automated screening process to routinely test 10 chiral supercritical fluid chromatography (SFC) methods - five columns combined with two co-solvents - as part of a chiral separation lab workflow. Proprietary software tools enabled automated method screening of racemates, parallel evaluation of the resulting chromatograms for enantiomer separation and report generation. This process is largely automated and resulted in an efficient and reliable lab process with a minimum requirement for human intervention. Screenings were conducted on a test set of 756 racemates that were selected with focus on structural variation and on 2667 proprietary samples from lab routines. Statistical analysis revealed that up to 92% of the tested racemic mixtures could be successfully separated with at least one of the tested conditions of the screening. Process efficiency was further increased by identification of optimal method screening sequence, re-definition of the optimal column set and project-specific adaptations considering reduced structural variation of the analytes. This study illustrates the usefulness of consistent chromatographic data sets to accelerate and facilitate the identification of chiral methods to separate enantiomers by automated processing and statistical analysis.

Evolution of Novartis' Small Molecule Screening Deck Design.

This article summarizes the evolution of the screening deck at the Novartis Institutes for BioMedical Research (NIBR). Historically, the screening deck was an assembly of all available compounds. In 2015, we designed a first deck to facilitate access to diverse subsets with optimized properties. We allocated the compounds as plated subsets on a 2D grid with property based ranking in one dimension and increasing structural redundancy in the other. The learnings from the 2015 screening deck were applied to the design of a next generation in 2019. We found that using traditional leadlikeness criteria (mainly MW, clogP) reduces the hit rates of attractive chemical starting points in subset screening. Consequently, the 2019 deck relies on solubility and permeability to select preferred compounds. The 2019 design also uses NIBR's experimental assay data and inferred biological activity profiles in addition to structural diversity to define redundancy across the compound sets.

Lymphocyte predominant cells detect Moraxella catarrhalis-derived antigens in nodular lymphocyte-predominant Hodgkin lymphoma.

Nodular lymphocyte-predominant Hodgkin lymphoma (NLPHL) is a rare lymphoma of B-cell origin with frequent expression of functional B-cell receptors (BCRs). Here we report that expression cloning followed by antigen screening identifies DNA-directed RNA polymerase beta' (RpoC) from Moraxella catarrhalis as frequent antigen of BCRs of IgD+ LP cells. Patients show predominance of HLA-DRB1*04/07 and the IgVH genes encode extraordinarily long CDR3s. High-titer, light-chain-restricted anti-RpoC IgG1/κ-type serum-antibodies are additionally found in these patients. RpoC and MID/hag, a superantigen co-expressed by Moraxella catarrhalis that is known to activate IgD+ B cells by binding to the Fc domain of IgD, have additive activation effects on the BCR, the NF-κB pathway and the proliferation of IgD+ DEV cells expressing RpoC-specific BCRs. This suggests an additive antigenic and superantigenic stimulation of B cells with RpoC-specific IgD+ BCRs under conditions of a permissive MHC-II haplotype as a model of NLPHL lymphomagenesis, implying future treatment strategies.

Reducing the Concepts of Data Science and Machine Learning to Tools for the Bench Chemist.

Machine Learning and Data Science have enjoyed a renaissance due to the availability of increased computational power and larger data sets. Many questions can be now asked and answered, that previously were beyond our scope. This does not translate instantly into new tools that can be used by those not skilled in the field, as many of the issues and traps still exist. In this paper, we look at some of the new tools that we have created, and some of the difficulties that still need to be taken care of during the transition from a project run by an expert, to a tool for the bench chemist.

Cell adhesion-induced transient interaction of ADAM15 with poly(A) binding protein at the cell membrane colocalizes with mRNA translation.

The regulation of temporo-spatial compartmentalization of protein synthesis is of crucial importance for a variety of physiologic cellular functions. Here, we demonstrate that the cell membrane-anchored disintegrin metalloproteinase ADAM15, upregulated in a variety of aggressively growing tumor cells, in the hyperproliferative synovial membrane of inflamed joints as well as in osteoarthritic chondrocytes, transiently binds to poly(A) binding protein 1 (PABP) in cells undergoing adhesion. The cytoplasmic domain of ADAM15 was shown to selectively interact with the proline-rich linker of PABP. Immunostainings of adhesion-triggered cells demonstrate an ADAM15-dependent recruitment of PABP to cell membrane foci coinciding with ongoing mRNA translation as visualized by the detection of puromycin-terminated polypeptides. Moreover, the increase in cell membrane-associated neosynthesis of puromycylated proteins upon induction of cell adhesion was proven linked to ADAM15 expression in HeLa and ADAM15-transfected chondrocytic cells. Thus, down regulation of ADAM15 by siRNA and/or the use of a cell line transfected with a mutant ADAM15-construct lacking the cytoplasmic tail resulted in a considerable reduction in the amount of cell membrane-associated puromycylated proteins formed during induced cell adhesion. These results provide first direct evidence for a regulatory role of ADAM15 on mRNA translation at the cell membrane that transiently emerges in response to triggering cell adhesion and might have potential implications under pathologic conditions of matrix remodeling associated with ADAM15 upregulation.

Chiral Cliffs: Investigating the Influence of Chirality on Binding Affinity.

Chirality is understood by many as a binary concept: a molecule is either chiral or it is not. In terms of the action of a structure on polarized light, this is indeed true. When examined through the prism of molecular recognition, the answer becomes more nuanced. In this work, we investigated chiral behavior on protein-ligand binding: when does chirality make a difference in binding activity? Chirality is a property of the 3D structure, so recognition also requires an appreciation of the conformation. In many situations, the bioactive conformation is undefined. We set out to address this by defining and using several novel 2D descriptors to capture general characteristic features of the chiral center. Using machine-learning methods, we built different predictive models to estimate if a chiral pair (a set of two enantiomers) might exhibit a chiral cliff in a binding assay. A set of about 3800 chiral pairs extracted from the ChEMBL23 database was used to train and test our models. By achieving an accuracy of up to 75 %, our models provide good performance in discriminating chiral cliffs from non-cliffs. More importantly, we were able to derive some simple guidelines for when one can reasonably use a racemate and when an enantiopure compound is needed in an assay. We critically discuss our results and show detailed examples of using our guidelines. Along with this publication we provide our dataset, our novel descriptors, and the Python code to rebuild the predictive models.

Chemical Topic Modeling: Exploring Molecular Data Sets Using a Common Text-Mining Approach.

Big data is one of the key transformative factors which increasingly influences all aspects of modern life. Although this transformation brings vast opportunities it also generates novel challenges, not the least of which is organizing and searching this data deluge. The field of medicinal chemistry is not different: more and more data are being generated, for instance, by technologies such as DNA encoded libraries, peptide libraries, text mining of large literature corpora, and new in silico enumeration methods. Handling those huge sets of molecules effectively is quite challenging and requires compromises that often come at the expense of the interpretability of the results. In order to find an intuitive and meaningful approach to organizing large molecular data sets, we adopted a probabilistic framework called "topic modeling" from the text-mining field. Here we present the first chemistry-related implementation of this method, which allows large molecule sets to be assigned to "chemical topics" and investigating the relationships between those. In this first study, we thoroughly evaluate this novel method in different experiments and discuss both its disadvantages and advantages. We show very promising results in reproducing human-assigned concepts using the approach to identify and retrieve chemical series from sets of molecules. We have also created an intuitive visualization of the chemical topics output by the algorithm. This is a huge benefit compared to other unsupervised machine-learning methods, like clustering, which are commonly used to group sets of molecules. Finally, we applied the new method to the 1.6 million molecules of the ChEMBL22 data set to test its robustness and efficiency. In about 1 h we built a 100-topic model of this large data set in which we could identify interesting topics like "proteins", "DNA", or "steroids". Along with this publication we provide our data sets and an open-source implementation of the new method (CheTo) which will be part of an upcoming version of the open-source cheminformatics toolkit RDKit.

Anti-citrullinated protein antibodies cause arthritis by cross-reactivity to joint cartilage.

Today, it is known that autoimmune diseases start a long time before clinical symptoms appear. Anti-citrullinated protein antibodies (ACPAs) appear many years before the clinical onset of rheumatoid arthritis (RA). However, it is still unclear if and how ACPAs are arthritogenic. To better understand the molecular basis of pathogenicity of ACPAs, we investigated autoantibodies reactive against the C1 epitope of collagen type II (CII) and its citrullinated variants. We found that these antibodies are commonly occurring in RA. A mAb (ACC1) against citrullinated C1 was found to cross-react with several noncitrullinated epitopes on native CII, causing proteoglycan depletion of cartilage and severe arthritis in mice. Structural studies by X-ray crystallography showed that such recognition is governed by a shared structural motif "RG-TG" within all the epitopes, including electrostatic potential-controlled citrulline specificity. Overall, we have demonstrated a molecular mechanism that explains how ACPAs trigger arthritis.

Prediction of protein mutation effects based on dehydration and hydrogen bonding - A large-scale study.

Reliable computational prediction of protein side chain conformations and the energetic impact of amino acid mutations are the key aspects for the optimization of biotechnologically relevant enzymatic reactions using structure-based design. By improving the protein stability, higher yields can be achieved. In addition, tuning the substrate selectivity of an enzymatic reaction by directed mutagenesis can lead to higher turnover rates. This work presents a novel approach to predict the conformation of a side chain mutation along with the energetic effect on the protein structure. The HYDE scoring concept applied here describes the molecular interactions primarily by evaluating the effect of dehydration and hydrogen bonding on molecular structures in aqueous solution. Here, we evaluate its capability of side-chain conformation prediction in classic remutation experiments. Furthermore, we present a new data set for evaluating "cross-mutations," a new experiment that resembles real-world application scenarios more closely. This data set consists of protein pairs with up to five point mutations. Thus, structural changes are attributed to point mutations only. In the cross-mutation experiment, the original protein structure is mutated with the aim to predict the structure of the side chain as in the paired mutated structure. The comparison of side chain conformation prediction ("remutation") showed that the performance of HYDEprotein is qualitatively comparable to state-of-the art methods. The ability of HYDEprotein to predict the energetic effect of a mutation is evaluated in the third experiment. Herein, the effect on protein stability is predicted correctly in 70% of the evaluated cases. Proteins 2017; 85:1550-1566. © 2017 Wiley Periodicals, Inc.