ADMET-AI: a machine learning ADMET platform for evaluation of large-scale chemical libraries.

MOTIVATION: The emergence of large chemical repositories and combinatorial chemical spaces, coupled with high-throughput docking and generative AI, have greatly expanded the chemical diversity of small molecules for drug discovery. Selecting compounds for experimental validation requires filtering these molecules based on favourable druglike properties, such as Absorption, Distribution, Metabolism, Excretion, and Toxicity (ADMET). RESULTS: We developed ADMET-AI, a machine learning platform that provides fast and accurate ADMET predictions both as a website and as a Python package. ADMET-AI has the highest average rank on the TDC ADMET Leaderboard, and it is currently the fastest web-based ADMET predictor, with a 45% reduction in time compared to the next fastest public ADMET web server. ADMET-AI can also be run locally with predictions for one million molecules taking just 3.1 h. AVAILABILITY AND IMPLEMENTATION: The ADMET-AI platform is freely available both as a web server at admet.ai.greenstonebio.com and as an open-source Python package for local batch prediction at github.com/swansonk14/admet_ai (also archived on Zenodo at doi.org/10.5281/zenodo.10372930). All data and models are archived on Zenodo at doi.org/10.5281/zenodo.10372418.

Identifying digenic disease genes via machine learning in the Undiagnosed Diseases Network.

Rare diseases affect millions of people worldwide, and discovering their genetic causes is challenging. More than half of the individuals analyzed by the Undiagnosed Diseases Network (UDN) remain undiagnosed. The central hypothesis of this work is that many of these rare genetic disorders are caused by multiple variants in more than one gene. However, given the large number of variants in each individual genome, experimentally evaluating combinations of variants for potential to cause disease is currently infeasible. To address this challenge, we developed the digenic predictor (DiGePred), a random forest classifier for identifying candidate digenic disease gene pairs by features derived from biological networks, genomics, evolutionary history, and functional annotations. We trained the DiGePred classifier by using DIDA, the largest available database of known digenic-disease-causing gene pairs, and several sets of non-digenic gene pairs, including variant pairs derived from unaffected relatives of UDN individuals. DiGePred achieved high precision and recall in cross-validation and on a held-out test set (PR area under the curve > 77%), and we further demonstrate its utility by using digenic pairs from the recent literature. In contrast to other approaches, DiGePred also appropriately controls the number of false positives when applied in realistic clinical settings. Finally, to enable the rapid screening of variant gene pairs for digenic disease potential, we freely provide the predictions of DiGePred on all human gene pairs. Our work enables the discovery of genetic causes for rare non-monogenic diseases by providing a means to rapidly evaluate variant gene pairs for the potential to cause digenic disease.

Genetics of agenesis/hypoplasia of the uterus and vagina: narrowing down the number of candidate genes for Mayer-Rokitansky-Küster-Hauser Syndrome.

PURPOSE: Mayer-Rokitansky-Küster-Hauser (MRKH) syndrome consists of congenital absence of the uterus and vagina and is often associated with renal, skeletal, cardiac, and auditory defects. The genetic basis is largely unknown except for rare variants in several genes. Many candidate genes have been suggested by mouse models and human studies. The purpose of this study was to narrow down the number of candidate genes. METHODS: Whole exome sequencing was performed on 111 unrelated individuals with MRKH; variant analysis focused on 72 genes suggested by mouse models, human studies of physiological candidates, or located near translocation breakpoints in t(3;16). Candidate variants (CV) predicted to be deleterious were confirmed by Sanger sequencing. RESULTS: Sanger sequencing verified 54 heterozygous CV from genes identified through mouse (13 CV in 6 genes), human (22 CV in seven genes), and translocation breakpoint (19 CV in 11 genes) studies. Twelve patients had ≥ 2 CVs, including four patients with two variants in the same gene. One likely digenic combination of LAMC1 and MMP14 was identified. CONCLUSION: We narrowed 72 candidate genes to 10 genes that appear more likely implicated. These candidate genes will require further investigation to elucidate their role in the development of MRKH.

Molecular mechanisms and structural features of cardiomyopathy-causing troponin T mutants in the tropomyosin overlap region.

Point mutations in genes encoding sarcomeric proteins are the leading cause of inherited primary cardiomyopathies. Among them are mutations in the TNNT2 gene that encodes cardiac troponin T (TnT). These mutations are clustered in the tropomyosin (Tm) binding region of TnT, TNT1 (residues 80-180). To understand the mechanistic changes caused by pathogenic mutations in the TNT1 region, six hypertrophic cardiomyopathy (HCM) and two dilated cardiomyopathy (DCM) mutants were studied by biochemical approaches. Binding assays in the absence and presence of actin revealed changes in the affinity of some, but not all, TnT mutants for Tm relative to WT TnT. HCM mutants were hypersensitive and DCM mutants were hyposensitive to Ca2+ in regulated actomyosin ATPase activities. To gain better insight into the disease mechanism, we modeled the structure of TNT1 and its interactions with Tm. The stability predictions made by the model correlated well with the affinity changes observed in vitro of TnT mutants for Tm. The changes in Ca2+ sensitivity showed a strong correlation with the changes in binding affinity. We suggest the primary reason by which these TNNT2 mutations between residues 92 and 144 cause cardiomyopathy is by changing the affinity of TnT for Tm within the TNT1 region.

Mechanistic heterogeneity in contractile properties of α-tropomyosin (TPM1) mutants associated with inherited cardiomyopathies.

The most frequent known causes of primary cardiomyopathies are mutations in the genes encoding sarcomeric proteins. Among those are 30 single-residue mutations in TPM1, the gene encoding α-tropomyosin. We examined seven mutant tropomyosins, E62Q, D84N, I172T, L185R, S215L, D230N, and M281T, that were chosen based on their clinical severity and locations along the molecule. The goal of our study was to determine how the biochemical characteristics of each of these mutant proteins are altered, which in turn could provide a structural rationale for treatment of the cardiomyopathies they produce. Measurements of Ca(2+) sensitivity of human ß-cardiac myosin ATPase activity are consistent with the hypothesis that hypertrophic cardiomyopathies are hypersensitive to Ca(2+) activation, and dilated cardiomyopathies are hyposensitive. We also report correlations between ATPase activity at maximum Ca(2+) concentrations and conformational changes in TnC measured using a fluorescent probe, which provide evidence that different substitutions perturb the structure of the regulatory complex in different ways. Moreover, we observed changes in protein stability and protein-protein interactions in these mutants. Our results suggest multiple mechanistic pathways to hypertrophic and dilated cardiomyopathies. Finally, we examined a computationally designed mutant, E181K, that is hypersensitive, confirming predictions derived from in silico structural analysis.

Generation of two induced pluripotent stem cell lines from patients with Williams syndrome.

Williams syndrome (WS) is a relatively rare genetic disorder. It arises from a microdeletion in chromosome 7q11.23, resulting in the loss of one copy of more than 20 genes. Disorders in multiple systems, including cardiovascular and nervous systems, occur in patients with WS. Here, we generated two human induced pluripotent stem cell (iPSC) lines from WS patients. Both lines expressed pluripotency markers at gene and protein levels. They possessed normal karyotypes and the potential to differentiate into three germ layers. They serve as a useful tool to study disease mechanism, test drugs, and identify promising therapeutics for patients with WS.

ADMET-AI: A machine learning ADMET platform for evaluation of large-scale chemical libraries.

Summary: The emergence of large chemical repositories and combinatorial chemical spaces, coupled with high-throughput docking and generative AI, have greatly expanded the chemical diversity of small molecules for drug discovery. Selecting compounds for experimental validation requires filtering these molecules based on favourable druglike properties, such as Absorption, Distribution, Metabolism, Excretion, and Toxicity (ADMET). We developed ADMET-AI, a machine learning platform that provides fast and accurate ADMET predictions both as a website and as a Python package. ADMET-AI has the highest average rank on the TDC ADMET Benchmark Group leaderboard, and it is currently the fastest web-based ADMET predictor, with a 45% reduction in time compared to the next fastest ADMET web server. ADMET-AI can also be run locally with predictions for one million molecules taking just 3.1 hours. Availability and Implementation: The ADMET-AI platform is freely available both as a web server at admet.ai.greenstonebio.com and as an open-source Python package for local batch prediction at github.com/swansonk14/admet_ai (also archived on Zenodo at doi.org/10.5281/zenodo.10372930 ). All data and models are archived on Zenodo at doi.org/10.5281/zenodo.10372418 .

SGLT2 inhibitor ameliorates endothelial dysfunction associated with the common ALDH2 alcohol flushing variant.

The common aldehyde dehydrogenase 2 (ALDH2) alcohol flushing variant known as ALDH2*2 affects ∼8% of the world's population. Even in heterozygous carriers, this missense variant leads to a severe loss of ALDH2 enzymatic activity and has been linked to an increased risk of coronary artery disease (CAD). Endothelial cell (EC) dysfunction plays a determining role in all stages of CAD pathogenesis, including early-onset CAD. However, the contribution of ALDH2*2 to EC dysfunction and its relation to CAD are not fully understood. In a large genome-wide association study (GWAS) from Biobank Japan, ALDH2*2 was found to be one of the strongest single-nucleotide polymorphisms associated with CAD. Clinical assessment of endothelial function showed that human participants carrying ALDH2*2 exhibited impaired vasodilation after light alcohol drinking. Using human induced pluripotent stem cell-derived ECs (iPSC-ECs) and CRISPR-Cas9-corrected ALDH2*2 iPSC-ECs, we modeled ALDH2*2-induced EC dysfunction in vitro, demonstrating an increase in oxidative stress and inflammatory markers and a decrease in nitric oxide (NO) production and tube formation capacity, which was further exacerbated by ethanol exposure. We subsequently found that sodium-glucose cotransporter 2 inhibitors (SGLT2i) such as empagliflozin mitigated ALDH2*2-associated EC dysfunction. Studies in ALDH2*2 knock-in mice further demonstrated that empagliflozin attenuated ALDH2*2-mediated vascular dysfunction in vivo. Mechanistically, empagliflozin inhibited Na+/H+-exchanger 1 (NHE-1) and activated AKT kinase and endothelial NO synthase (eNOS) pathways to ameliorate ALDH2*2-induced EC dysfunction. Together, our results suggest that ALDH2*2 induces EC dysfunction and that SGLT2i may potentially be used as a preventative measure against CAD for ALDH2*2 carriers.

Personalized structural biology reveals the molecular mechanisms underlying heterogeneous epileptic phenotypes caused by de novo KCNC2 variants.

Whole-exome sequencing (WES) in the clinic has identified several rare monogenic developmental and epileptic encephalopathies (DEE) caused by ion channel variants. However, WES often fails to provide actionable insight for rare diseases, such as DEEs, due to the challenges of interpreting variants of unknown significance (VUS). Here, we describe a "personalized structural biology" (PSB) approach that leverages recent innovations in the analysis of protein 3D structures to address this challenge. We illustrate this approach in an Undiagnosed Diseases Network (UDN) individual with DEE symptoms and a de novo VUS in KCNC2 (p.V469L), the Kv3.2 voltage-gated potassium channel. A nearby KCNC2 variant (p.V471L) was recently suggested to cause DEE-like phenotypes. Computational structural modeling suggests that both affect protein function. However, despite their proximity, the p.V469L variant is likely to sterically block the channel pore, while the p.V471L variant is likely to stabilize the open state. Biochemical and electrophysiological analyses demonstrate heterogeneous loss-of-function and gain-of-function effects, as well as differential response to 4-aminopyridine treatment. Molecular dynamics simulations illustrate that the pore of the p.V469L variant is more constricted, increasing the energetic barrier for K+ permeation, whereas the p.V471L variant stabilizes the open conformation. Our results implicate variants in KCNC2 as causative for DEE and guide the interpretation of a UDN individual. They further delineate the molecular basis for the heterogeneous clinical phenotypes resulting from two proximal pathogenic variants. This demonstrates how the PSB approach can provide an analytical framework for individualized hypothesis-driven interpretation of protein-coding VUS.

Nonmulberry Silk Fibroin Scaffold Shows Superior Osteoconductivity Than Mulberry Silk Fibroin in Calvarial Bone Regeneration.

Recent years have witnessed the advancement of silk biomaterials in bone tissue engineering, although clinical application of the same is still in its infancy. In this study, the potential of pure nonmulberry Antheraea mylitta (Am) fibroin scaffold, without preloading with bone precursor cells, to repair calvarial bone defect in a rat model is explored and compared with its mulberry counterpart Bombyx mori (Bm) silk fibroin. After 3 months of implantation, Am scaffold culminates in a completely ossified regeneration with a progressive increase in mineralization at the implanted site. On the other hand, the Bm scaffold fails to repair the damaged bone, presumably due to its low osteoconductivity and early degradation. The deposition of bone matrix on scaffolds is evaluated by scanning electron and atomic force microscopy. These results are corroborated by in vitro studies of enzymatic degradation, colony formation, and secondary conformational features of the scaffold materials. The greater biocompatibility and mineralization in pure nonmulberry fibroin scaffolds warrants the use of these scaffolds as an "ideal bone graft" biomaterial for effective repair of critical size defects.