Applications of Machine Learning in Drug Target Discovery

Chronic postsurgical inguinal pain: incidence and diagnostic biomarkers from a large German national claims database

BACKGROUND

Chronic postsurgical inguinal pain (CPIP) is the most common complication of groin hernia surgery. The characteristics of patients, their medical care, and choice of diagnostic tools remain to be defined to optimise preventive and therapeutic interventions.

METHODS

Claims data from 2018 and a 1-yr follow-up were analysed for incidence and medical care. A separate cohort (141 healthy controls and 17 CPIP patients) was examined by deep phenotyping. This included sensory testing, blood and skin biopsies, MRI of the dorsal root ganglion (DRG), and patient-reported outcomes.

RESULTS

Of 11,221 patients with hernia surgery in 2018 identified, 8.5% had pain before that was relieved by surgery, but a similar percentage had novel pain in this region. Deep phenotyping of 141 healthy controls provided a map of the inguinal sensory system. The following analysis of patients with CPIP revealed that they suffered from moderate pain with neuropathic features, individual sensory abnormalities, and unilateral L1 DRG atrophy. In the blood, levels of C-C-motif chemokine ligand (CCL2) and brain-derived neurotrophic factor (BDNF) were upregulated, whereas apolipoprotein A1 (ApoA1) concentration was reduced. A cluster of DRG atrophy, BDNF, ApoA1, and anxiety correlated best with the diagnosis. CPIP patients with novel pain had significantly more DRG atrophy (-24% ipsilateral vs contralateral volume).

CONCLUSIONS

CPIP is often newly acquired after surgery. A combination of DRG imaging, serum markers, and anxiety screening can support the diagnosis. In the future, this could guide clinicians towards more personalised therapies (e.g. targeting anxiety or lipid profiles) and possible altered surgical techniques.

CLINICAL TRIAL REGISTRATION

German Trial Registry DRKS00024588 and DRKS00016790.

Evolution of Drug Development and Regulatory Affairs: The Demonstrated Power of Artificial Intelligence

PURPOSE

Artificial intelligence (AI) refers to technology capable of mimicking human cognitive functions and has important applications across all sectors and industries, including drug development. This has considerable implications for the regulation of drug development processes, as it is expected to transform both the way drugs are brought to market and the systems through which this process is controlled. There is currently insufficient evidence in published literature of the real-world applications of AI. Therefore, this narrative review investigated, collated, and elucidated the applications of AI in drug development and its regulatory processes.

METHODS

A narrative review was conducted to ascertain the role of AI in streamlining drug development and regulatory processes.

FINDINGS

The findings of this review revealed that machine learning or deep learning, natural language processing, and robotic process automation were favored applications of AI. Each of them had considerable implications on the operations they were intended to support. Overall, the AI tools facilitated access and provided manageability of information for decision-making across the drug development lifecycle. However, the findings also indicate that additional work is required by regulatory authorities to set out appropriate guidance on applications of the technology, which has critical implications for safety, regulatory process workflow and product development costs.

IMPLICATIONS

AI has adequately proven its utility in drug development, prompting further investigations into the translational value of its utility based on cost and time saved for the delivery of essential drugs.

Optimized models and deep learning methods for drug response prediction in cancer treatments: a review

Recent advancements in deep learning (DL) have played a crucial role in aiding experts to develop personalized healthcare services, particularly in drug response prediction (DRP) for cancer patients. The DL's techniques contribution to this field is significant, and they have proven indispensable in the medical field. This review aims to analyze the diverse effectiveness of various DL models in making these predictions, drawing on research published from 2017 to 2023. We utilized the VOS-Viewer 1.6.18 software to create a word cloud from the titles and abstracts of the selected studies. This study offers insights into the focus areas within DL models used for drug response. The word cloud revealed a strong link between certain keywords and grouped themes, highlighting terms such as deep learning, machine learning, precision medicine, precision oncology, drug response prediction, and personalized medicine. In order to achieve an advance in DRP using DL, the researchers need to work on enhancing the models' generalizability and interoperability. It is also crucial to develop models that not only accurately represent various architectures but also simplify these architectures, balancing the complexity with the predictive capabilities. In the future, researchers should try to combine methods that make DL models easier to understand; this will make DRP reviews more open and help doctors trust the decisions made by DL models in cancer DRP.

Comparative Studies on Resampling Techniques in Machine Learning and Deep Learning Models for Drug-Target Interaction Prediction

The prediction of drug-target interactions (DTIs) is a vital step in drug discovery. The success of machine learning and deep learning methods in accurately predicting DTIs plays a huge role in drug discovery. However, when dealing with learning algorithms, the datasets used are usually highly dimensional and extremely imbalanced. To solve this issue, the dataset must be resampled accordingly. In this paper, we have compared several data resampling techniques to overcome class imbalance in machine learning methods as well as to study the effectiveness of deep learning methods in overcoming class imbalance in DTI prediction in terms of binary classification using ten (10) cancer-related activity classes from BindingDB. It is found that the use of Random Undersampling (RUS) in predicting DTIs severely affects the performance of a model, especially when the dataset is highly imbalanced, thus, rendering RUS unreliable. It is also found that SVM-SMOTE can be used as a go-to resampling method when paired with the Random Forest and Gaussian Naïve Bayes classifiers, whereby a high F1 score is recorded for all activity classes that are severely and moderately imbalanced. Additionally, the deep learning method called Multilayer Perceptron recorded high F1 scores for all activity classes even when no resampling method was applied.

In silico Methods for Identification of Potential Therapeutic Targets

At the initial stage of drug discovery, identifying novel targets with maximal efficacy and minimal side effects can improve the success rate and portfolio value of drug discovery projects while simultaneously reducing cycle time and cost. However, harnessing the full potential of big data to narrow the range of plausible targets through existing computational methods remains a key issue in this field. This paper reviews two categories of in silico methods—comparative genomics and network-based methods—for finding potential therapeutic targets among cellular functions based on understanding their related biological processes. In addition to describing the principles, databases, software, and applications, we discuss some recent studies and prospects of the methods. While comparative genomics is mostly applied to infectious diseases, network-based methods can be applied to infectious and non-infectious diseases. Nonetheless, the methods often complement each other in their advantages and disadvantages. The information reported here guides toward improving the application of big data-driven computational methods for therapeutic target discovery.

Graphical abstract

Multiple Poses and Thermodynamics of Ligands Targeting Protein Surfaces: The Case of Furosemide Binding to mitoNEET in Aqueous Solution

Human NEET proteins, such as NAF-1 and mitoNEET, are homodimeric, redox iron-sulfur proteins characterized by triple cysteine and one histidine-coordinated [2Fe-2S] cluster. They exist in an oxidized and reduced state. Abnormal release of the cluster is implicated in a variety of diseases, including cancer and neurodegeneration. The computer-aided and structure-based design of ligands affecting cluster release is of paramount importance from a pharmaceutical perspective. Unfortunately, experimental structural information so far is limited to only one ligand/protein complex. This is the X-ray structure of furosemide bound to oxidized mitoNEET. Here we employ an enhanced sampling approach, Localized Volume-based Metadynamics, developed by some of us, to identify binding poses of furosemide to human mitoNEET protein in solution. The binding modes show a high variability within the same shallow binding pocket on the protein surface identified in the X-ray structure. Among the different binding conformations, one of them is in agreement with the crystal structure’s one. This conformation might have been overstabilized in the latter because of the presence of crystal packing interactions, absent in solution. The calculated binding affinity is compatible with experimental data. Our protocol can be used in a straightforward manner in drug design campaigns targeting this pharmaceutically important family of proteins.

GPCRs Are Optimal Regulators of Complex Biological Systems and Orchestrate the Interface between Health and Disease

GPCRs arguably represent the most effective current therapeutic targets for a plethora of diseases. GPCRs also possess a pivotal role in the regulation of the physiological balance between healthy and pathological conditions; thus, their importance in systems biology cannot be underestimated. The molecular diversity of GPCR signaling systems is likely to be closely associated with disease-associated changes in organismal tissue complexity and compartmentalization, thus enabling a nuanced GPCR-based capacity to interdict multiple disease pathomechanisms at a systemic level. GPCRs have been long considered as controllers of communication between tissues and cells. This communication involves the ligand-mediated control of cell surface receptors that then direct their stimuli to impact cell physiology. Given the tremendous success of GPCRs as therapeutic targets, considerable focus has been placed on the ability of these therapeutics to modulate diseases by acting at cell surface receptors. In the past decade, however, attention has focused upon how stable multiprotein GPCR superstructures, termed receptorsomes, both at the cell surface membrane and in the intracellular domain dictate and condition long-term GPCR activities associated with the regulation of protein expression patterns, cellular stress responses and DNA integrity management. The ability of these receptorsomes (often in the absence of typical cell surface ligands) to control complex cellular activities implicates them as key controllers of the functional balance between health and disease. A greater understanding of this function of GPCRs is likely to significantly augment our ability to further employ these proteins in a multitude of diseases.

Artificial Intelligence for Autonomous Molecular Design: A Perspective

Domain-aware artificial intelligence has been increasingly adopted in recent years to expedite molecular design in various applications, including drug design and discovery. Recent advances in areas such as physics-informed machine learning and reasoning, software engineering, high-end hardware development, and computing infrastructures are providing opportunities to build scalable and explainable AI molecular discovery systems. This could improve a design hypothesis through feedback analysis, data integration that can provide a basis for the introduction of end-to-end automation for compound discovery and optimization, and enable more intelligent searches of chemical space. Several state-of-the-art ML architectures are predominantly and independently used for predicting the properties of small molecules, their high throughput synthesis, and screening, iteratively identifying and optimizing lead therapeutic candidates. However, such deep learning and ML approaches also raise considerable conceptual, technical, scalability, and end-to-end error quantification challenges, as well as skepticism about the current AI hype to build automated tools. To this end, synergistically and intelligently using these individual components along with robust quantum physics-based molecular representation and data generation tools in a closed-loop holds enormous promise for accelerated therapeutic design to critically analyze the opportunities and challenges for their more widespread application. This article aims to identify the most recent technology and breakthrough achieved by each of the components and discusses how such autonomous AI and ML workflows can be integrated to radically accelerate the protein target or disease model-based probe design that can be iteratively validated experimentally. Taken together, this could significantly reduce the timeline for end-to-end therapeutic discovery and optimization upon the arrival of any novel zoonotic transmission event. Our article serves as a guide for medicinal, computational chemistry and biology, analytical chemistry, and the ML community to practice autonomous molecular design in precision medicine and drug discovery.

Data-Driven Discovery of Molecular Targets for Antibody-Drug Conjugates in Cancer Treatment

Antibody-drug conjugate therapy has attracted considerable attention in recent years. Since the selection of appropriate targets is a critical aspect of antibody-drug conjugate research and development, a big data research for discovery of candidate targets per tumor type is outstanding and of high interest. Thus, the purpose of this study was to identify and prioritize candidate antibody-drug conjugate targets with translational potential across common types of cancer by mining the Human Protein Atlas, as a unique big data resource. To perform a multifaceted screening process, XML and TSV files including immunohistochemistry expression data for 45 normal tissues and 20 tumor types were downloaded from the Human Protein Atlas website. For genes without high protein expression across critical normal tissues, a quasi H-score (range, 0-300) was computed per tumor type. All genes with a quasi H - score ≥ 150 were extracted. Of these, genes with cell surface localization were selected and included in a multilevel validation process. Among 19670 genes that encode proteins, 5520 membrane protein-coding genes were included in this study. During a multistep data mining procedure, 332 potential targets were identified based on the level of the protein expression across critical normal tissues and 20 tumor types. After validation, 23 cell surface proteins were identified and prioritized as candidate antibody-drug conjugate targets of which two have interestingly been approved by the FDA for use in solid tumors, one has been approved for lymphoma, and four have currently been entered in clinical trials. In conclusion, we identified and prioritized several candidate targets with translational potential, which may yield new clinically effective and safe antibody-drug conjugates. This large-scale antibody-based proteomic study allows us to go beyond the RNA-seq studies, facilitates bench-to-clinic research of targeted anticancer therapeutics, and offers valuable insights into the development of new antibody-drug conjugates.

Machine Learning Methods in Drug Discovery

The advancements of information technology and related processing techniques have created a fertile base for progress in many scientific fields and industries. In the fields of drug discovery and development, machine learning techniques have been used for the development of novel drug candidates. The methods for designing drug targets and novel drug discovery now routinely combine machine learning and deep learning algorithms to enhance the efficiency, efficacy, and quality of developed outputs. The generation and incorporation of big data, through technologies such as high-throughput screening and high through-put computational analysis of databases used for both lead and target discovery, has increased the reliability of the machine learning and deep learning incorporated techniques. The use of these virtual screening and encompassing online information has also been highlighted in developing lead synthesis pathways. In this review, machine learning and deep learning algorithms utilized in drug discovery and associated techniques will be discussed. The applications that produce promising results and methods will be reviewed.