Search

Summary: Deeply understanding the properties (e.g., chemical or biological characteristics) of small molecules plays an essential role in drug development. A large number of molecular property datasets have been rapidly accumulated in recent years. However, most of these datasets contain only a limited amount of data, which hinders deep learning methods from making accurate predictions of the corresponding molecular properties. In this work, we propose a transfer learning strategy to alleviate such a data scarcity problem by exploiting the similarity between molecular property prediction tasks. We introduce an effective and interpretable computational framework, named MoTSE (Molecular Tasks Similarity Estimator), to provide an accurate estimation of task similarity. Comprehensive tests demonstrated that the task similarity derived from MoTSE can serve as useful guidance to improve the prediction performance of transfer learning on molecular properties. We also showed that MoTSE can capture the intrinsic relationships between molecular properties and provide meaningful interpretability for the derived similarity.

Peptide-protein interactions play fundamental roles in cellular processes and are crucial for designing peptide therapeutics. Here, the authors present a deep learning framework for simultaneously predicting peptide-protein interactions and identifying peptide binding residues involved in the interactions.

The IDG-DREAM Challenge carried out crowdsourced benchmarking of predictive algorithms for kinase inhibitor activities on unpublished data. This study provides a resource to compare emerging algorithms and prioritize new kinase activities to accelerate drug discovery and repurposing efforts.

Abstract The global spread of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) requires an urgent need to find effective therapeutics for the treatment of coronavirus disease 2019 (COVID-19). In this study, we developed an integrative drug repositioning framework, which fully takes advantage of machine learning and statistical analysis approaches to systematically integrate and mine large-scale knowledge graph, literature and transcriptome data to discover the potential drug candidates against SARS-CoV-2. Our in silico screening followed by wet-lab validation indicated that a poly-ADP-ribose polymerase 1 (PARP1) inhibitor, CVL218, currently in Phase I clinical trial, may be repurposed to treat COVID-19. Our in vitro assays revealed that CVL218 can exhibit effective inhibitory activity against SARS-CoV-2 replication without obvious cytopathic effect. In addition, we showed that CVL218 can interact with the nucleocapsid (N) protein of SARS-CoV-2 and is able to suppress the LPS-induced production of several inflammatory cytokines that are highly relevant to the prevention of immunopathology induced by SARS-CoV-2 infection.

Accurate identification of compound–protein interactions (CPIs) in silico may deepen our understanding of the underlying mechanisms of drug action and thus remarkably facilitate drug discovery and development. Conventional similarity- or docking-based computational methods for predicting CPIs rarely exploit latent features from currently available large-scale unlabeled compound and protein data and often limit their usage to relatively small-scale datasets. In the present study, we propose DeepCPI, a novel general and scalable computational framework that combines effective feature embedding (a technique of representation learning) with powerful deep learning methods to accurately predict CPIs at a large scale. DeepCPI automatically learns the implicit yet expressive low-dimensional features of compounds and proteins from a massive amount of unlabeled data. Evaluations of the measured CPIs in large-scale databases, such as ChEMBL and BindingDB, as well as of the known drug–target interactions from DrugBank, demonstrated the superior predictive performance of DeepCPI. Furthermore, several interactions among small-molecule compounds and three G protein-coupled receptor targets (glucagon-like peptide-1 receptor, glucagon receptor, and vasoactive intestinal peptide receptor) predicted using DeepCPI were experimentally validated. The present study suggests that DeepCPI is a useful and powerful tool for drug discovery and repositioning. The source code of DeepCPI can be downloaded from https://github.com/FangpingWan/DeepCPI. Keywords: Deep learning, Machine learning, Drug discovery, In silico drug screening, Compound–protein interaction prediction

Synthetic lethality (SL), an important type of genetic interaction, can provide useful insight into the target identification process for the development of anticancer therapeutics. Although several well-established SL gene pairs have been verified to be conserved in humans, most SL interactions remain cell-line specific. Here, we demonstrated that the cell-line-specific gene expression profiles derived from the shRNA perturbation experiments performed in the LINCS L1000 project can provide useful features for predicting SL interactions in human. In this paper, we developed a semi-supervised neural network-based method called EXP2SL to accurately identify SL interactions from the L1000 gene expression profiles. Through a systematic evaluation on the SL datasets of three different cell lines, we demonstrated that our model achieved better performance than the baseline methods and verified the effectiveness of using the L1000 gene expression features and the semi-supervise training technique in SL prediction.

To enhance the service life of agitator blades in phosphoric acid reactors, the NiCoCrMo0.2 medium-entropy alloy (Mo0.2 MEA) was designed using the cluster-plus-glue-atom model. The microstructure and erosion resistance of Mo0.2 MEA were analyzed, revealing a composition of FCC solid-solution dendrites with small μ phases. Mo0.2 MEA exhibits a higher microhardness (256.4 HV0.2) compared to its Mo-free counterpart. Additionally, the lowest Icorr value and largest impedance radius indicate superior corrosion resistance relative to 904L steel, attributed to a dense and stable passive film of MoO3 and Cr2O3. The erosion resistance of Mo0.2 MEA, assessed by cumulative mass loss, is double that of 904L stainless steel. Furthermore, the Stacking-fault energy of Mo0 MEA (∼−105.7 mJ/m2) is lower than that of Mo0.2 MEA (∼−85.91 mJ/m2) and 904L (∼251.2 mJ/m2), suggesting greater plastic deformation capability in Mo0 MEA. These findings offer a new approach for developing anti-erosion medium-entropy alloys through high-entropy alloying and laser additive manufacturing, guided by the cluster-plus-glue-atom model.

Biodegradable film is considered a promising alternative to conventional plastic film in agriculture production. Differences in degradation rates result in varying effects on soil temperature and moisture, which directly affect crop growth and yield. However, studies on the effects of biodegradable films with different degradation rates on crop growth remain limited. To investigate these effects, a field experiment was conducted in 2019 and 2020, featuring three biodegradable films with degradation induction periods of 30 days (M1), 60 days (M2) and 90 days (M3), and a non-mulching control (CK). The results indicated that the degradation rates of the three films followed the expected order of M1 > M2 > M3, with final breakage rates of 27.23 %, 23.68 %, and 2.73 % in 2019, and 38.28 %, 28.63 %, and 7.39 % in 2020, respectively. Biodegradable film mulching increased average soil moisture, temperature, and the content of NO3–-N and NH4+-N throughout the entire maize growth period. Due to its fastest degradation rate, M1 exhibited weaker warming and moisture-retention effects compared to M2 and M3. The favorable soil conditions created by biodegradable film mulching promoted maize growth, advanced the peak times of plant height and leaf area index, and increased maize yield. Compared to CK, M1, M2, and M3 increased maize yield by 12.96 %, 14.84 %, and 15.86 % in 2019, and 15.12 %, 16.29 %, and 15.91 % in 2020, respectively. Furthermore, biodegradable film mulching also increased maize water use efficiency and nitrogen partial factor productivity by reducing soil evaporation and increasing maize yield. The EWM-TOPSIS model ranked M2 as the optimal treatment for both years, followed by M1 and M3. This study provides valuable reference for determining biodegradable films with suitable degradation rates in the experimental region.

Objective: This study aimed to investigate the current status of affiliated stigma and caregiver burden among parents of children with epilepsy, analyze their correlation, and identify factors influencing affiliated stigma. Methods: A cross-sectional survey was conducted among 194 parents of children with epilepsy who met the inclusion and exclusion criteria in Shenzhen City, Guangdong Province, China. Data were collected through questionnaires, including a demographic information sheet, an affiliated stigma scale, and a caregiver burden scale. Results: The results revealed that parents of children with epilepsy experienced a moderate level of affiliated stigma, with an average score of 54.92 ± 10.44. Similarly, caregiver burden scores fell within the moderate range, with an average score of 44.14 ± 16.02. Factors influencing affiliated stigma scores included the frequency of epileptic seizures in children, the types of anti-epileptic medications taken by children, and the place of residence. The total caregiver burden score and scores in various dimensions (emotional, cognitive, and behavioral) of caregivers for epilepsy patients were positively correlated with the affiliated stigma score. Affiliated stigma was found to independently explain 21.3 % of the variation in caregiver burden. Conclusion: In the future, healthcare professionals should develop targeted interventions for children with epilepsy and their parents to reduce affiliated stigma, decrease caregiver burden, and enhance the caregiving capabilities of parents of children with epilepsy. These measures are essential to improve the overall well-being of both parents and children affected by epilepsy.

With the development of highly integrated electronic equipment, electromagnetic pollution and heat congregation have become serious problems. Electromagnetic interference (EMI) shielding materials with high thermal conductivity and mechanical properties are highly desirable. Herein, the reduced graphene oxide/Ni nanowire aerogels (GNiA) were firstly prepared via hydrothermal reducing process and freeze-drying. Subsequently, using epoxy (EP) as a polymer matrix, the GNiA/EP composites with high compression strength of 147 MPa were fabricated by vacuum filtration and curing. The GNiA endows the composites with great electromagnetic wave loss, as well as improved phonon propagation efficiency, resulting in the super high EMI SE. With the GNiA content of 12.14 wt%, the EMI SE and thermal conductivity of GNiA/EP composites can reach as high as 85 dB and 1.69 W/(m·K) respectively. The obtained GNiA/EP composites provide a strategy to meet the demand for electronic equipment for new-generation EMI shielding materials.

RNA polymerase II (Pol II) generally pauses at certain positions along gene bodies, thereby interrupting the transcription elongation process, which is often coupled with various important biological functions, such as precursor mRNA splicing and gene expression regulation. Characterizing the transcriptional elongation dynamics can thus help us understand many essential biological processes in eukaryotic cells. However, experimentally measuring Pol II elongation rates is generally time and resource consuming. We developed PEPMAN (polymerase II elongation pausing modeling through attention-based deep neural network), a deep learning-based model that accurately predicts Pol II pausing sites based on the native elongating transcript sequencing (NET-seq) data. Through fully taking advantage of the attention mechanism, PEPMAN is able to decipher important sequence features underlying Pol II pausing. More importantly, we demonstrated that the analyses of the PEPMAN-predicted results around various types of alternative splicing sites can provide useful clues into understanding the cotranscriptional splicing events. In addition, associating the PEPMAN prediction results with different epigenetic features can help reveal important factors related to the transcription elongation process. All these results demonstrated that PEPMAN can provide a useful and effective tool for modeling transcription elongation and understanding the related biological factors from available high-throughput sequencing data. [ABSTRACT FROM AUTHOR]

High-entropy intermetallic compounds (HEIs) with high strength and outstanding wear resistance are possibly the desirable coating materials. In this present work, A FeCoNiAl HEI composition originated from the multi-component substitution of A site in L12-Ni3Al was designed. Next, the HEI coating was achieved using laser cladding. The Cu atoms were incorporated into the HEI coatings to investigate their effects on the phase structure, cracking, and properties of HEI coatings. The results show that the FeCoNiAl HEI coating comprises a sole BCC phase. Adding Cu atoms less than 4.76 at.% does not alter the phase structures. Besides, the study found that as x exceeded 0.3, many tiny rod-like secondary precipitates are generated in the B2 matrix. The study also found that the FeCoNiAl HEI coating hardness at approximately 550 HV0.2 is twice that of Ni3Al coating. The secondary precipitates in the Cu-rich HEI coatings enhance the hardness to about 700 HV0.2. Accordingly, the wear resistance shows a similar tendency to the microhardness. Moreover, the Cu0.4 HEI coating exhibited the lowest friction coefficient (∼0.381) and the narrowest wear scar. In addition, the cracking behavior transforms from transgranular to intergranular cracking due to increased Cu content resulting in decreased cracking sensitivity. The study found that the corrosion resistance of Cux HEI coatings is more defective than those of Ni3Al coating. The Cu atoms effectively decrease the passivation current density of Cux HEI coatings. The above results provide a novel thought for the development of high-performance coatings.

Developing high-performance high-entropy alloys (HEAs) fabricated by laser additive manufacturing (LAM) is the pursuit of the metallic community. In the present work, we designed a series of [(Al6-xNbx)-(FeCoNi)12]Cr3 HEA compositions using a high-entropy alloying strategy based on a cluster-plus-glue-atom model. And their thin-wall-sharped bulks were fabricated by LAM and post-aging treatment. The effects of cracking inhibition and microstructure evolution on the tensile properties were researched in detail. The results show that as the Nb substitutes for Al atoms, the cracking behaviour is ameliorated, ascribed to the tiny Laves phase refined the dendrite spacings and back-filled in the inter-dendritic liquid film. Also, introducing Nb atoms improves the strength but deteriorates the ductility. Significantly, the Nb4 HEA possesses the best tensile-property combination (i.e. σs ∼ 419.2 MPa, σb ∼ 787.4 MPa, and δ ∼ 15.5%) with a strain mechanism of dislocation slip mode. After post-aging for 72 h, the microstructure comprises fully recrystallized equiaxed FCC grains and many tiny needle-like D019 precipitates, leading to high strength and sufficient ductility (i.e. σ0.2 ∼ 535.9 MPa, σb ∼820 MPa and δ value of 8.9%). These findings provide a new paradigm for the LAM of crack-free HEAs with excellent mechanical properties.

Two novel HEA compositions of [Nb¯-(FeCoNi)12]Cr3 (Nb¯ = Nb, V) were designed using a high-entropy alloying strategy by analyzing traditional IN718 superalloy based on a cluster-plus-glue-atom model. And their thin-wall-shaped bulks were prepared by laser additive manufacturing, with an emphasis on the effects of using V atoms to substitute for Nb atoms on their microstructures and mechanical properties. Also, the strengthening mechanism induced by D019 precipitates of as-aged V2Nb4 HEA was discussed. After using V to substitute for Nb, the contents of inter-dendritic C14-Laves phases are effectively inhibited. The ductility is significantly improved with a small sacrifice of strength (i.e., σb ∼ 851 MPa and δ ∼ 16.8 % for V2Nb4 HEA). After aging treatment, the main microstructure transforms from fine dendrites to recrystallized equiaxed grains. The intercrossed needle-shaped D019 precipitates with a volume fraction of 7.2 % are uniformly distributed in the FCC matrix, which hinders the dislocation slip. Thereto, the as-96-hr-aged V2Nb4 HEA exhibits the most excellent strength-ductility trade-off (e.g., σs ∼ 721.5 MPa, σb ∼ 979.5 MPa with a sufficient δ of 7.5 %). The above findings provide references for the development of LAM high-performance HEAs originating from traditional alloys.

The agitator blade made by 904 L stainless steel was subjected to serious corrosion-wear due to its low hardness in phosphoric acid reactors. In this aspect, various nano-TiC reinforced [Cr–Fe4Co4Ni4]Cr3 HEA composite coatings were designed using the cluster-plus-glue-atom model and fabricated on 904 L stainless steel by laser cladding. The phase structure of the composite coatings was composed of FCC solid solution and TiC phase. The microstructure observation detected that tiny TiC particles widely distributed along the inter-dendrites of FCC matrix. Also, the particle dimensions and volumes rapidly enhanced with the addition of TiC. Whereas, excessive TiC contents (≥12.5vol%) led to the generation of microcracks. The TEM results further confirmed that TiC particles did not decompose during laser cladding. With the addition of TiC, the microhardness, wear and corrosion resistance of the composite coatings gradually increased. Especially, the microhardness of [Cr–Fe4Co4Ni4]Cr3-15vol%TiC composite coating reached the peak value of 357.4 HV0.2, approximately twice higher than that of the substrate. It's specific wear rate (3.974 mm3 N−1 m−1) was lower than that of the substrate (5.545 mm3 N−1 m−1). Compared with 904 L stainless steel, the corrosion current density of [Cr–Fe4Co4Ni4]Cr3-15vol%TiC composite coating reduced by nearly an order of magnitude, the impedance increased by 3.5 times.

To introduce D022 superlattice (noted as γ′′ phase) precipitation strengthening in additive manufacturing, a design strategy of combining the overall valence electron concentration with the calculation of phase diagrams is proposed, and Ni2.1CoCrFe0.5Nb0.2 HEA is designed. The wall-shaped samples were prepared by AM, and after solution at 1100 °C for 2 h and aging at 650 °C for 120 h, the γ′′ phase with volume of 14% causes yield strength increase by 727 MPa, the yield strength increased dramatically to ∼1005 MPa, the ultimate strength increased dramatically to ∼1240 MPa, the tensile elongation maintained at ∼20%. The high strength results from the precipitation strengthening of the γ′′ phase, and the large ductility are primarily attributed to an evolution of multiple stacking fault structures. The present study will not only promote the development of high-performance HEAs by AM but also provides a pathway for achievement of AM technology industrial applications.

Vagus nerve stimulation (VNS) is a neuromodulation therapy increasingly used for treating drug-resistant epilepsy. However, it remains to be determined which patients are best suited for the treatment, and it is difficult to predict the therapeutic effect before the implantation. Mutations in some genes could lead to epilepsy. Here we report two cases of pediatric patients with drug-resistant epilepsy treated by VNS therapy: Patient 1 with ARX mutation achieved good outcomes; Patient 2 with the CDKL5 mutation did not show improvement. Additionally, the therapeutic impact of VNS on brain networks was investigated, hoping to provide some empirical evidence for a better understanding of the mechanism of VNS treatment.

2D ZnO is one of the most attractive materials for potential applications in photocatalysis, gas and light detection, ultraviolet light-emitting diodes, resistive memory, and pressure-sensitive devices. The electronic structures, magnetic properties, and optical properties of M (Li, Na, Mg, Ca, or Ga) and TM (Cr, Co, Cu, Ag, or Au) adsorbed g-ZnO were investigated with density functional theory (DFT). It is found that the band structure, charge density difference, electron spin density, work function, and absorption spectrum of g-ZnO can be tuned by adsorbing M or TM atoms. More specifically, the specific charge transfer occurs between g-ZnO and adsorbed atom, indicating the formation of a covalent bond. The work functions of M adsorbed g-ZnO systems are obviously smaller than that of intrinsic g-ZnO, implying great potential in high-efficiency field emission devices. The Li, Na, Mg, Ca, Ga, Ag, or Au adsorbed g-ZnO systems, the Cr adsorbed g-ZnO system, and the Co or Cu adsorbed g-ZnO systems exhibit non-magnetic semiconductor proprieties, magnetic semiconductor proprieties, and magnetic metal proprieties, respectively. In addition, the magnetic moments of Cr, Co, or Cu adsorbed g-ZnO systems are 4 μB, 3 μB, or 1 μB, respectively, which are mainly derived from adsorbed atoms, suggesting potential applications in nano-scale spintronics devices. Compared with the TM absorbed g-ZnO systems, the M adsorbed g-ZnO systems have more obvious absorption peaks for visible light, particularly for Mg or Ca adsorbed g-ZnO systems. Their absorption peaks appear in the near-infrared region, suggesting great potential in solar photocatalysis. Our work contributes to the design and fabrication of high-efficiency field emission devices, nano-scale spintronics devices, and visible-light responsive photocatalytic materials.

Using first principles based on density functional theory (DFT), the CO, NH3, NO, and NO2 gas adsorbed on intrinsic Graphite-like ZnO (g−ZnO) and vacancy-deficient g−ZnO were systematically studied. For intrinsic g−ZnO, the adsorption energy of NH3, NO, and NO2 adsorption defective g−ZnO systems increased significantly due to the introduction of Zn vacancy (VZn). Especially, for NH3, NO, and NO2 adsorbed Zn-vacancy g−ZnO (VZn/g−ZnO) systems increased to 1.366 eV, 2.540 eV and 2.532 eV, respectively. In addition, with the introduction of vacancies, the adsorption height of the gases adsorbed on VZn/g−ZnO system is significantly reduced, especially the adsorption height of the NH3 adsorbed on VZn/g−ZnO system is reduced to 0.686 Å. It is worth mentioning that the introduction of O-vacancy (VO) significantly enhances the charge transfer between NO or NO2 and VO/g−ZnO. This suggest that the defective g−ZnO is more suitable for detecting NH3, NO and NO2 gas. It is interesting to note that the adsorption of NO and NO2 gases gives rise to magnetic moments of 1 μB and 0.858 μB for g−ZnO, and 1 μB and 1 μB for VO/g−ZnO. In addition, VZn induced 1.996 μB magnetic moments for intrinsic g−ZnO, and the CO, NH3, NO and NO2 change the magnetic of VZn/g−ZnO. The adsorption of NO2 causes the intrinsic g−ZnO to exhibit metallic properties, while the adsorption of NH3 gas molecules causes VZn/g−ZnO also to show metallic properties. The adsorption of NO and NO2 causes VZn/g−ZnO to display semi-metallic properties. These results facilitate the enrichment of defect detection means and the design of gas detection devices.

Molecular doping is an excellent instrument to modify the electronic properties of two−dimensional materials. In our work, the structure and electronic properties of the adsorption systems of g−ZnO adsorbed by organic molecules (including Tetracyanoethylene (TCNE), Tetracyanoquinodimethane (TCNQ), and Tetrahydrofulvalene (TTF)) were investigated computationally using Density Functional Theory (DFT). The results showed that the TCNE and TCNQ, as electron receptors, doped the LUMO energy level above the valence band maximum (VBM) of the g−ZnO band structure, demonstrating effective p−type doping. The n−type doping of g−ZnO was obtained that the TTF molecules, as electron donors, doped the HOMO energy level below the conduction band minimum (CBM) of the band structure for g−ZnO. In addition, the TCNE, TCNQ, and TTF breathed additional holes or electrons into the monolayer g−ZnO, creating surface dipole moments between the g−ZnO and organic molecules, which caused work function to be adjustable, ranging from 3.871 eV to 5.260 eV. Our results prove that organic molecular doping was instrumental in improving the performance of g−ZnO−based nano−electronic devices, providing theoretical support for the fabrication of p−doping or n−doping nano−semiconductor components. The tunable range of field emission capability of g−ZnO−based electronic devices was also extended.

Previous studies have documented that people tend to respond faster and memorize better to the in-group traits. It may be particularly manifest for ethnic minorities, due to their salient ethnic identity. However, few studies have explored how the valence of traits modulates the in-group preference effect. The present study examined the impacts of ethnic identity salience and the valence of traits on the group-preference effect among 33 Han Chinese in a Tibetan-dominant area and 32 Tibetan participants in a Han-dominant area. Two weeks before the experiment, we measured the ethnic identity salience of participants in both groups. In the formal experiment, we used the group-reference effect (GRE) paradigm with three encoding tasks. The results showed that, regardless of whether ethnic identity was salient, both groups responded faster to positive traits than to negative traits when evaluating their own group, whereas there were no significant difference between the processing of positive traits and negative traits in the out-group evaluation and font judgment tasks. This suggested a pervasive processing advantage of the in-group positive characteristics. The results imply that self-enhancement motivation had a moderation effect on the GRE, as well as the ethnic identity salience may not be necessary for a GRE.

Based on the public’s cognition, evaluation and expectation of rainstorm warning information in Beijing, the descriptive statistics and non-parametric test methods were used for data analysis. The results show that more than 80% of the public can recognize the importance of rainstorm warning, but only half of the public pay attention to the rainstorm warning information. As for the method of issuing early warning, the village broadcasting and electronic display, and network communication channels need to be strengthened. Although the rainstorm warning information was released in time, the update process was not effectively communicated, and the information update period was long, it could not meet the public’s demand for real-time attention to heavy rain. As far as the content of the warning information is concerned, false alarm effects of the rainstorm warning exist. According to the results, it is concluded that lead time can be extended by adding the update of warning information, and the probability of rainfall could be introduced to rainstorm warning information. These guide the design of rainstorm warning and promote understanding how the public responds to the warning.

To explore whether the alteration of lncRNA expression is correlated with polycyclic aromatic hydrocarbons (PAHs) exposure and DNA damage, we examined PAHs external and internal exposure, DNA damage and lncRNAs (HOTAIR, MALAT1, TUG1 and GAS5) expression in peripheral blood lymphocytes (PBLCs) of 150 male coke oven workers and 60 non-PAHs exposure workers. We found the expression of HOTAIR, MALAT1, and TUG1 were enhanced in PBLCs of coke oven workers and positively correlated with the levels of external PAHs exposure (adjusted Ptrend

Regulatory agencies may, whether outside of set rules or within their discretion, depart from the original goals or principles set for enforcing the rules, which we term selective enforcement. Taking China, a country in transition, as an example, and using cases and large-sample tests, we present empirical evidence of selective enforcement. The results show that the China Securities Regulatory Commission (CSRC) takes into account whether companies violating the rules have a state-owned background and the strength of that background when investigating and punishing non-compliance. After controlling for the degree of violation, state-owned-enterprises (SOEs) are punished less severely than private companies; and the higher the hierarchy of the SOE in question, the less severe the punishment. It also takes longer for SOEs to be punished. We also find that companies that violate the rules less seriously have a greater tendency to apply for refinancing than those that violate the rules more seriously. This may be because the severity of the violation can affect listed companies’ expectations of obtaining refinancing. The analysis and conclusions of this study prove useful in understanding the causes and consequences of selective enforcement in transition economies.

Artificial intelligence (AI) and machine learning (ML) models are being deployed in many domains of society and have recently reached the field of drug discovery. Given the increasing prevalence of antimicrobial resistance, as well as the challenges intrinsic to antibiotic development, there is an urgent need to accelerate the design of new antimicrobial therapies. Antimicrobial peptides (AMPs) are therapeutic agents for treating bacterial infections, but their translation into the clinic has been slow owing to toxicity, poor stability, limited cellular penetration and high cost, among other issues. Recent advances in AI and ML have led to breakthroughs in our abilities to predict biomolecular properties and structures and to generate new molecules. The ML-based modelling of peptides may overcome some of the disadvantages associated with traditional drug discovery and aid the rapid development and translation of AMPs. Here, we provide an introduction to this emerging field and survey ML approaches that can be used to address issues currently hindering AMP development. We also outline important limitations that can be addressed for the broader adoption of AMPs in clinical practice, as well as new opportunities in data-driven peptide design., Competing Interests: Competing interests J.J.C. is scientific co-founder and scientific advisory board chair of EnBiotix, an antibiotic drug discovery company, and Phare Bio, a non-profit venture focused on antibiotic drug development. C.d.l.F.-N. provides consulting services to Invaio Sciences and is a member of the Scientific Advisory Boards of Nowture S.L. and Phare Bio. The remaining authors declare no competing interests.