Human-in-the-loop assisted de novo molecular design.

A de novo molecular design workflow can be used together with technologies such as reinforcement learning to navigate the chemical space. A bottleneck in the workflow that remains to be solved is how to integrate human feedback in the exploration of the chemical space to optimize molecules. A human drug designer still needs to design the goal, expressed as a scoring function for the molecules that captures the designer's implicit knowledge about the optimization task. Little support for this task exists and, consequently, a chemist usually resorts to iteratively building the objective function of multi-parameter optimization (MPO) in de novo design. We propose a principled approach to use human-in-the-loop machine learning to help the chemist to adapt the MPO scoring function to better match their goal. An advantage is that the method can learn the scoring function directly from the user's feedback while they browse the output of the molecule generator, instead of the current manual tuning of the scoring function with trial and error. The proposed method uses a probabilistic model that captures the user's idea and uncertainty about the scoring function, and it uses active learning to interact with the user. We present two case studies for this: In the first use-case, the parameters of an MPO are learned, and in the second use-case a non-parametric component of the scoring function to capture human domain knowledge is developed. The results show the effectiveness of the methods in two simulated example cases with an oracle, achieving significant improvement in less than 200 feedback queries, for the goals of a high QED score and identifying potent molecules for the DRD2 receptor, respectively. We further demonstrate the performance gains with a medicinal chemist interacting with the system.

LibINVENT: Reaction-based Generative Scaffold Decoration for in Silico Library Design.

Because of the strong relationship between the desired molecular activity and its structural core, the screening of focused, core-sharing chemical libraries is a key step in lead optimization. Despite the plethora of current research focused on in silico methods for molecule generation, to our knowledge, no tool capable of designing such libraries has been proposed. In this work, we present a novel tool for de novo drug design called LibINVENT. It is capable of rapidly proposing chemical libraries of compounds sharing the same core while maximizing a range of desirable properties. To further help the process of designing focused libraries, the user can list specific chemical reactions that can be used for the library creation. LibINVENT is therefore a flexible tool for generating virtual chemical libraries for lead optimization in a broad range of scenarios. Additionally, the shared core ensures that the compounds in the library are similar, possess desirable properties, and can also be synthesized under the same or similar conditions. The LibINVENT code is freely available in our public repository at https://github.com/MolecularAI/Lib-INVENT. The code necessary for data preprocessing is further available at: https://github.com/MolecularAI/Lib-INVENT-dataset.

Has Artificial Intelligence Impacted Drug Discovery?

Artificial intelligence (AI) tools find increasing application in drug discovery supporting every stage of the Design-Make-Test-Analyse (DMTA) cycle. The main focus of this chapter is the application in molecular generation with the aid of deep neural networks (DNN). We present a historical overview of the main advances in the field. We analyze the concepts of distribution and goal-directed learning and then highlight some of the recent applications of generative models in drug design with a focus into research work from the biopharmaceutical industry. We present in some more detail REINVENT which is an open-source software developed within our group in AstraZeneca and the main platform for AI molecular design support for a number of medicinal chemistry projects in the company and we also demonstrate some of our work in library design. Finally, we present some of the main challenges in the application of AI in Drug Discovery and different approaches to respond to these challenges which define areas for current and future work.

DockStream: a docking wrapper to enhance de novo molecular design.

Recently, we have released the de novo design platform REINVENT in version 2.0. This improved and extended iteration supports far more features and scoring function components, which allows bespoke and tailor-made protocols to maximize impact in small molecule drug discovery projects. A major obstacle of generative models is producing active compounds, in which predictive (QSAR) models have been applied to enrich target activity. However, QSAR models are inherently limited by their applicability domains. To overcome these limitations, we introduce a structure-based scoring component for REINVENT. DockStream is a flexible, stand-alone molecular docking wrapper that provides access to a collection of ligand embedders and docking backends. Using the benchmarking and analysis workflow provided in DockStream, execution and subsequent analysis of a variety of docking configurations can be automated. Docking algorithms vary greatly in performance depending on the target and the benchmarking and analysis workflow provides a streamlined solution to identifying productive docking configurations. We show that an informative docking configuration can inform the REINVENT agent to optimize towards improving docking scores using public data. With docking activated, REINVENT is able to retain key interactions in the binding site, discard molecules which do not fit the binding cavity, harness unused (sub-)pockets, and improve overall performance in the scaffold-hopping scenario. The code is freely available at https://github.com/MolecularAI/DockStream .

Discovery of Novel UDP-N-Acetylglucosamine Acyltransferase (LpxA) Inhibitors with Activity against Pseudomonas aeruginosa.

This study describes a novel series of UDP-N-acetylglucosamine acyltransferase (LpxA) inhibitors that was identified through affinity-mediated selection from a DNA-encoded compound library. The original hit was a selective inhibitor of Pseudomonas aeruginosa LpxA with no activity against Escherichia coli LpxA. The biochemical potency of the series was optimized through an X-ray crystallography-supported medicinal chemistry program, resulting in compounds with nanomolar activity against P. aeruginosa LpxA (best half-maximal inhibitory concentration (IC50) <5 nM) and cellular activity against P. aeruginosa (best minimal inhibitory concentration (MIC) of 4 µg/mL). Lack of activity against E. coli was maintained (IC50 > 20 µM and MIC > 128 µg/mL). The mode of action of analogues was confirmed through genetic analyses. As expected, compounds were active against multidrug-resistant isolates. Further optimization of pharmacokinetics is needed before efficacy studies in mouse infection models can be attempted. To our knowledge, this is the first reported LpxA inhibitor series with selective activity against P. aeruginosa.

De novo design with deep generative models based on 3D similarity scoring.

We have demonstrated the utility of a 3D shape and pharmacophore similarity scoring component in molecular design with a deep generative model trained with reinforcement learning. Using Dopamine receptor type 2 (DRD2) as an example and its antagonist haloperidol 1 as a starting point in a ligand based design context, we have shown in a retrospective study that a 3D similarity enabled generative model can discover new leads in the absence of any other information. It can be efficiently used for scaffold hopping and generation of novel series. 3D similarity based models were compared against 2D QSAR based, indicating a significant degree of orthogonality of the generated outputs and with the former having a more diverse output. In addition, when the two scoring components are combined together for training of the generative model, it results in more efficient exploration of desirable chemical space compared to the individual components.

Interlimb Asymmetries: The Need for an Individual Approach to Data Analysis.

ABSTRACT: Bishop, C, Lake, J, Loturco, I, Papadopoulos, K, Turner, A, and Read, P. Interlimb asymmetries: the need for an individual approach to data analysis. J Strength Cond Res 35(3): 695-701, 2021-It has been shown that the magnitude of interlimb asymmetries varies depending on the test selected; however, literature relating to whether asymmetries always favor the same limb is scarce. The aim of this study was to determine whether interlimb asymmetries always favored the same side for common metrics across unilateral strength and jumping-based tests. Twenty-eight recreational sport athletes performed unilateral isometric squats, single-leg countermovement jumps, and single-leg broad jumps with asymmetries in peak force compared across all tests, and eccentric and concentric impulse asymmetries compared between jumps. Mean asymmetries for all tests were low (≤-5.3%), and all interlimb differences for jump tests favored the left limb, whereas asymmetries during the isometric squat favored the right limb. Despite the low mean asymmetry values, individual data highlighted substantially greater differences. Levels of agreement for asymmetries were computed through the Kappa coefficient and ranged from slight to substantial (<0.01-0.79), although concentric impulse asymmetries for jump tests was the only comparison with result in substantial levels of agreement. With asymmetries rarely being present on the same side across tests, these results show that a more individual approach to reporting asymmetries is required, which should help practitioners when designing targeted training interventions for their reduction.

REINVENT 2.0: An AI Tool for De Novo Drug Design.

In the past few years, we have witnessed a renaissance of the field of molecular de novo drug design. The advancements in deep learning and artificial intelligence (AI) have triggered an avalanche of ideas on how to translate such techniques to a variety of domains including the field of drug design. A range of architectures have been devised to find the optimal way of generating chemical compounds by using either graph- or string (SMILES)-based representations. With this application note, we aim to offer the community a production-ready tool for de novo design, called REINVENT. It can be effectively applied on drug discovery projects that are striving to resolve either exploration or exploitation problems while navigating the chemical space. It can facilitate the idea generation process by bringing to the researcher's attention the most promising compounds. REINVENT's code is publicly available at https://github.com/MolecularAI/Reinvent.

Acute and chronic effects of foam rolling vs eccentric exercise on ROM and force output of the plantar flexors.

Foam rolling and eccentric exercise interventions have been demonstrated to improve range of motion (ROM). However, these two modalities have not been directly compared. Twenty-three academy soccer players (age: 18 ± 1; height: 1.74 ± 0.08 m; body mass: 69.3 ± 7.5 kg) were randomly allocated to either a foam rolling (FR) or eccentric exercise intervention designed to improve dorsiflexion ROM. Participants performed the intervention daily for a duration of four weeks. Measurements of dorsiflexion ROM, isometric plantar flexion torque and drop jump reactive strength index were taken at baseline (pre-intervention) and at three subsequent time-points (30-min post, 24-hours post and 4-weeks post). A significant time x group interaction effect was observed for dorsiflexion (P = 0.036), but not for torque or reactive strength index. For dorsiflexion, there was a significant increase in both acute (30-min; P < 0.001) and chronic (4-week; P < 0.001) ROM for the eccentric group, whilst FR exhibited only an acute improvement (P < 0.001). Eccentric training would appear a more efficacious modality than foam rolling for improving dorsiflexion ROM in elite academy soccer players.

TAVR facilitated by high-pressure balloon post-dilatation to fracture the ring of the small dysfunctional aortic Mosaic bioprosthesis.

Treatment of a failing aortic bioprosthesis by transcatheter valve-in-valve (ViV) therapy has become an alternative to redo surgery. However, the ViV technique may be less effective in small surgical valves because of patient/prosthesis mismatch (PPM). Here we will discuss the bioprosthetic valve fracture/remodelling (BVF) procedure and the most important issues regarding this promising new technique.

Renal injury following intravitreal anti-VEGF administration in diabetic patients with proliferative diabetic retinopathy and chronic kidney disease--a possible side effect?

The use of intravitreal injections of anti-Vascular Endothelial Growth Factor (anti-VEGF) has been used for a broad spectrum of ocular pathologic entities. Although the dose of anti-VEGF agents used for treating eye disease is minute compared with that used intravenously, intraocular administration can lead to systemic absorption and reduce serum VEGF levels. Several systemic side effects, such as hypertension and cardiovascular complications have been rarely reported in the literature. Renal complications of intravenous administration of anti-VEGF, are well known and include a variety of renal pathological damage which can induce proteinuria and hypertension. We describe herein, 2 cases of diabetic patients with preexisting kidney disease who presented severe reduction of their renal function after intraocular administration of anti-VEGF. Although a cause -effect correlation cannot be established unless further studies are performed, we believe that pretreatment counseling should include a discussion outlining the possible risk of aggravating of the renal function in patients with kidney disease. Close cooperation with the patient's nephrologist and close monitoring of the patient may be required, in such cases, in order to monitor the renal function before and after the intravitreal administration of anti-VEGF.

Association of estrogen receptor alpha (ERα) gene polymorphisms with endometrial thickness and lipid profile in women with breast cancer treated with aromatase inhibitors.

Aromatase inhibitors (AIs) provide an alternative to tamoxifen as an adjuvant therapy for post-menopausal, hormone-receptor positive breast cancer patients. The aim of the present study was to evaluate the effect of PvuII and XbaI polymorphisms of the ERα gene at ΑΙs treatment's adverse effects in post-menopausal women with breast cancer. The study included 87 post-menopausal women with ER-positive breast cancer treated with AIs and 80 healthy controls. The overall presence of ERα polymorphisms in all women with breast cancer was not different from the healthy controls. Endometrial thickness under AIs treatment was reduced from (mean value ± SD) 6,404 ± 2,901 mm to 3,666 ± 1,4656 mm. Moreover, the AA XbaI genotype was associated with greater reduction in endometrial thickness during therapy with AIs (p = 0.005). The presence of the CC PvuII and the AA XbaI genotypes were associated with elevated LDL levels and elevated triglycerides. In conclusion, the results of the present study showed that the genotype of women with breast cancer under AIs treatment might influence treatment's adverse effects, as, the presence of the CC PvuII and the AA XbaI genotypes of the ERα were associated with elevated LDL and triglycerides serum levels, while the AA XbaI genotype was associated with a greater reduction in endometrial thickness.

Hypermethylation-associated transcriptional silencing of E-cadherin in primary sporadic colorectal carcinomas.

Loss of E (epithelial)-cadherin expression has been previously documented in sporadic colorectal carcinomas (SCRCs), but not as a consequence of mutations or allelic loss. In this study, the methylation status of the E-cadherin promoter was examined by utilizing the methylation-specific polymerase chain reaction (MSP) assay in 63 primary SCRCs and paired adjacent normal tissues. This was correlated with E-cadherin expression at both the RNA and the protein levels using multiplex reverse transcription (RT)-PCR and immunohistochemistry (IHC), respectively. Data were associated with the patients' clinicopathological features. Methylated alleles were present in 34/61 (56%) of the samples examined. Decreased E-cadherin mRNA expression was demonstrated in 29/61 carcinomas (47.5%) and was significantly associated with lymph node (LN) metastases (p = 0.03, Kruskal-Wallis) and tumour stages Astler-Coller B1 and B2 (p = 0.01, chi(2)). E-cadherin IHC expression was significantly associated with the absence of LN metastases (p = 0.01, chi(2)) and tumour stages Astler-Coller B1 and B2 (p = 0.002, Kruskal-Wallis) in 28/63 (44.4%) of the samples examined. Twenty-three out of 29 (79.3%) samples with decreased mRNA expression and 20/33 (60.6%) with detected protein expression revealed methylated (p = 0.03, Kruskal-Wallis) and unmethylated (p = 0.01, Kruskal-Wallis) alleles, respectively. In agreement with previous work demonstrating that somatic mutations and loss of heterozygosity of the E-cadherin gene are rare or absent in the majority of SCRCs studied so far, this study reports a consistent and uniform decrease or absence of E-cadherin expression, associated with aberrant methylation, in the majority of carcinomas examined, suggesting an epigenetically mediated loss of E-cadherin function in these carcinomas.