Biology subjects

Yang, H.

Publications and source records attributed to Yang, H..

At least 415 records · Page 23Linked to original sources

bioRxiv2022

Nonlinear effect of light intensity on normal axial development of rhesus monkeys

PurposeTo investigate the effects of different indoor lighting intensity (500 lx, 750 lx and 1,000 lx) on normal ocular axial length growth by using juvenal rhesus monkeys. MethodsTwenty-four juvenile monkeys were exposed continuously to normal intensity light (NIL, 500 lx, n=16), medium intensity light (MIL, 750 lx, n=8) and high intensity light (HIL, 1 000 lx, n=8), with a same CCT value (about 3000 K) for 246 days. Axial length, anterior chamber depth, corneal curvature radius were measured at about a monthly interval. ResultsAfter 246 days of light exposure, the growth of axial length of the MIL group (750 lx) were 0.151 {+/-} 0.081 mm and 0.139 {+/-} 0.070 mm in the right and left eyes, respectively, and significantly larger in comparison with the NIL group (500lx, OD: 0.068 {+/-} 0.055 mm, OS: 0.074 {+/-} 0.057 mm) and the HIL group (1000lx,OD:0.063 {+/-} 0.093 mm, OS: 0.084 {+/-} 0.052 mm) monkeys. This effect was stable and robust during the whole experimental period. ConclusionThe effects of different intensity lighting on normal ocular axial development was not linear as most people currently think. We must be cautious when it comes to elevate light intensity in classrooms. Whether this conclusion is correct under lights of other CCT value needs further study.

developmental biology↗

bioRxiv2022

Accurate and efficient protein sequence designthrough learning concise local environment of residues

Protein sequence design has been widely applied in rational protein engineering and increasing the design accuracy and efficiency is highly desired. Here we present ProDESIGN-LE, an accurate and efficient design approach, which adopts a concise but informative representation of residues local environment and trains a transformer to select an appropriate residue at a position from its local environment. ProDESIGN-LE iteratively applies the transformer on the positions in the target structure, eventually acquiring a designed sequence with all residues fitting well with their local environments. ProDESIGN-LE designed sequences for 68 naturally occurring and 129 hallucinated proteins within 20 seconds per protein on average, and the predicted structures from the designed sequences perfectly resemble the target structures with state-of-the-art average TM-score exceeding 0.80. We further experimentally validated ProDESIGN-LE by designing five sequences for an enzyme, chloramphenicol O-acetyltransferase type III (CAT III), and recombinantly expressing the proteins in E. coli. Of these proteins, three exhibited excellent solubility, and one yielded monomeric species with circular dichroism spectra consistent with the natural CAT III protein.

bioinformatics↗

bioRxiv2022

A Novel Selective ERK1/2 Inhibitor, Laxiflorin B, Targets EGFR Mutation Subtypes in Non-small-cell Lung Cancer

Extracellular regulated protein kinases 1/2 (ERK1/2) are key members of multiple signaling pathways including the ErbB axis. ERK1/2 ectopic activation is responsible for various types of cancer, especially drug resistance to inhibitors of RTK, RAF and MEK, but specific ERK1/2 inhibitors are scarce. In this study, we identified a potential novel ERK covalent inhibitor, Laxiflorin B, a herbal compound with anticancer activity. However, Laxiflorin B is present at low levels in herbs; therefore, we adopted a semi-synthetic process for the efficient production of Laxiflorin B to improve the yield. Laxiflorin B induced mitochondria-mediated apoptosis via BAD activation in non-small-cell lung cancer (NSCLC) cells, especially in EGFR mutant subtypes. Transcriptomic analysis suggested that Laxiflorin B inhibits amphiregulin (AREG) and epiregulin (EREG) expression through ERK inhibition, and suppressed the activation of their receptors, ErbBs, via a positive feedback loop. More importantly, mass spectrometry combined with computer simulation analysis revealed that Laxiflorin B binds covalently to Cys-183 in the ATP-binding pocket of ERK1 through D-ring, and Cys-178 of ERK1 though non-inhibitory binding of A-ring, respectively. Laxiflorin B also exhibited strong tumor suppressive effects with low toxicity in a NSCLC tumor xenograft model in nude mice, and AREG and EREG were identified as biomarkers of Laxiflorin B efficacy. Finally, Laxiflorin B-4, a C-6 modification of Laxiflorin B, exhibited higher affinity for ERK1/2 and stronger tumor suppression. These findings provide a new approach to tumor inhibition using natural anticancer compounds.

pharmacology and toxicology↗

bioRxiv2022

An engineered xCas12i with high activity, high specificity and broad PAM range

The type-V CRISPR effector Cas12i, with its smaller size, short crRNA guiding, and self-processing features, is a potentially versatile genome editing tool. By screening Cas12i proteins from a metagenomic database, we identified a natural variant with high activity in mammalian cells, named as xCas12i. We further engineered the PAM-interacting, REC, and RuvC domains for enhanced cleavage activity and specificity. This variant, named as high-fidelity Cas12Max, exhibited robust genome editing activity and minimal off-target activity with a broad 5-TN recognition profile. With the fusion of deaminase TadA8e and further optimization of xCas12i, the base editor dCas12i-Tad8e also showed the high editing efficiency. This study provides highly efficient and specific tools for gene therapy.

bioengineering↗

bioRxiv2022

Surface anchoring of the Kingella kingae galactan is dependent on the lipopolysaccharide O-antigen

Kingella kingae is a leading cause of bone and joint infections and other invasive diseases in young children. A key K. kingae virulence determinant is a secreted exopolysaccharide that mediates resistance to serum complement and neutrophils and is required for full pathogenicity. The K. kingae exopolysaccharide is a galactofuranose homopolymer called galactan and is encoded by the pamABC genes in the pamABCDE locus. In this study, we sought to define the mechanism by which galactan is tethered on the bacterial surface, a prerequisite for mediating evasion of host immune mechanisms. We found that the pamD and pamE genes are glycosyltransferases and are required for synthesis of an atypical lipopolysaccharide (LPS) O-antigen. The LPS O-antigen in turn is required for anchoring of galactan, a novel mechanism for association of an exopolysaccharide with the bacterial surface. SignificanceKingella kingae is an emerging pediatric pathogen and produces invasive disease by colonizing the oropharynx, invading the bloodstream, and disseminating to distant sites. This organism produces a uniquely multifunctional exopolysaccharide called galactan that is critical for virulence and promotes intravascular survival by mediating resistance to serum and neutrophils. In this study, we established that at least some galactan is anchored to the bacterial surface via a novel structural interaction with an atypical lipopolysaccharide O-antigen. Additionally, we demonstrated that the atypical O-antigen is synthesized by the pamD and pamE genes, located downstream of the gene cluster responsible for galactan biosynthesis. This work addresses how the K. kingae exopolysaccharide can mediate innate immune resistance and advances understanding of bacterial exopolysaccharides and lipopolysaccharides.

microbiology↗

bioRxiv2022

Trans-Driver: a deep learning approach for cancer driver gene discovery with multi-omics data

Driver genes play a crucial role in the growth of cancer cells. Accurate identification of cancer driver genes is helping to strengthen the understanding of cancer pathogenesis and is conducive to the development of cancer treatment and drug-targe driver genes. However, due to the diversity and complexity of the multi-omics data, it is still challenging to identify cancer drivers.In this study, we propose Trans-Driver, a deep supervised learning method with a novel transformer network, which integrates multi-omics data to learn the differences and associations between different omics data for cancer drivers discovery. Compared with other state-of-the-art driver gene identification methods, Trans-Driver has achieved excellent performance on TCGA and CGC data Machine learning for multi-omics data integration in cancer. Among 20,000 protein-coding genes, Trans-Driver reported 185 candidate driver genes, of which 103 genes (about 55%) were included in the gold standard CGC data set. Finally, we analyzed the contribution of each feature to the identification of driver genes. We found that the integration of multi-omics data can improve the performance of our method compared with using only somatic mutation data. Through detailed analysis, we found that the candidate drivers are clinically meaningful, proving the practicability of Trans-Driver. Author summaryMany methods have been developed to identify cancer driver genes. However, most of these methods use single-omics data for cancer driver gene identification. Multi-omics-based methods for cancer driver gene identification are rare. Trans-Driver uses deep learning to process multi-omics data and learn the relationships between multi-omics data for cancer driver gene prediction. We have predicted 185 candidate cancer driver genes out of among 20,000 protein-coding genes. Also, we performed cancer driver gene prediction on 33 cancer types, and we identified the cancer driver genes corresponding to each cancer type. And, we observed that the predicted cancer driver genes were shown to have a role in cancer progression in recent studies. Our proposed method for cancer driver gene identification using multi-omics data has improved performance compared to using mutation data alone.

bioinformatics↗

bioRxiv2022

Sex differences in the nasal microbiome of healthy young adults

Respiratory diseases impose an immense health burden worldwide. Epidemiological studies have revealed extensive disparities in the incidence and severity of respiratory tract infections (RTIs) between males and females. It is recently hypothesized that there might also be a nasal microbiome axis contributing to the observed sex disparities, but without evidence. In this work, we study the nasal microbiome of healthy young adults in, as of today, the largest cohort based on deep shot-gun metagenomic sequencing. We mainly focus on the bacteriome, but also integrate the mycobiome to get a more holistic perspective. De novo assembly is performed to catalog the nasal bacterial colonizers/residents, which also identify and therefore account for uncharacterized components of the community. The bacteriome is then profiled based on the non-redundant metagenome-assembled genomes (MAGs) catalog constructed therefrom. Unsupervised clustering reveals clearly separable structural patterns in the nasal microbiome between the two sexes. Following this link, we systematically evaluate sex differences for the first time and reveal extensive sex-specific features in the nasal microbiome composition. More importantly, through network analyses, we capture markedly higher ecological stability and antagonistic potentials in the nasal microbiome of females than that of males. The analysis of the keystone bacteria of the communities reveal that the sex-dependent evolutionary characteristics might have contributed to this difference. HighlightsThe non-redundant nasal bacterial MAGs catalog constructed from ultra-deeply sequenced metagenomic data provides a valuable resource. Integrating nasal bacteriome and mycobiome data provides a more holistic perspective for the understudied human nasal microbiome. Unsupervised clustering helps uncover extensive sex differences in the nasal microbiome compositions. Network analyses capture markedly higher ecological stability and antagonistic potentials in the nasal microbiome of females than that of males. Sex-dependent genetic evolutionary forces play a role in the shaping of keystones in the nasal microbial community.

microbiology↗

bioRxiv2022

Fibroblasts-derived from Pluripotent Cells Harboring a Single Allele Knockout in Two Pluripotency Genes Exhibit DNA Methylation Abnormalities and pluripotency induction Defects

A complete knockout (KO) of a single key pluripotency gene has been shown to drastically affect embryonic stem cell (ESC) function and epigenetic reprogramming. However, knockin (KI)/KO of a reporter gene only in one of two alleles in a single pluripotency gene is considered harmless and is largely used in the stem cell field. Here, we sought to understand the impact of simultaneous elimination of a single allele in two ESC key genes on pluripotency potential and acquisition. We established multiple pluripotency systems harboring KI/KO in a single allele of two different pluripotency genes (i.e. Nanog+/-; Sall4+/-, Nanog+/-; Utf1+/-, Nanog+/-; Esrrb+/- and Sox2+/-; Sall4+/-). Interestingly, although these double heterozygous mutant lines maintain their stemness and contribute to chimeras equally to their parental control cells, fibroblasts derived from these systems show a significant reduction in their capability to induce pluripotency either by Oct4, Sox2, Klf4 and Myc (OSKM) or by nuclear transfer (NT). Tracing the expression of Sall4 and Nanog, as representative key pluripotency targeted genes, at early phases of reprogramming could not explain the seen delay/blockage. Further exploration identifies abnormal methylation landscape around pluripotent and developmental genes in the double heterozygous mutant fibroblasts. Accordingly, treatment with 5-azacytidine two days prior to transgene induction rescues the reprogramming defects. This study emphasizes the importance of maintaining two intact alleles for pluripotency induction and suggests that insufficient levels of key pluripotency genes leads to DNA methylation abnormalities in the derived-somatic cells later on in development.

cell biology↗

bioRxiv2022

Dendritic cell Piezo1 integrating mechanical stiffness and inflammatory signals directs the differentiation of TH1 and Treg cells in cancer

Dendritic cells (DCs) play an important role in anti-tumor immunity by inducing T cell differentiation. Herein, we found that the mechanical sensor Piezo1 expressed by DCs integrates innate inflammatory stimuli and stiffness signals and directs the reciprocal differentiation of TH1 and regulatory T (Treg) cells in cancer. Genetic deletion of Piezo1 in DCs inhibited the generation of TH1 cells while driving the development of Treg cells in promoting cancer growth. Mechanistically, Piezo1-deficient DCs regulated the secretion of the polarizing cytokines TGF{beta}1 and IL-12, leading to increased TGF{beta}R2-p-Smad3 activity and decreased IL-12R{beta}2-p-STAT4 activity while inducing the reciprocal differentiation of Treg and TH1 cells. In addition, Piezo1 integrated the SIRT1-hypoxia-inducible factor-1 alpha (HIF1)-dependent metabolic pathway and calcium-calcineurin-NFAT signaling pathway to orchestrate reciprocal TH1 and Treg lineage commitment through DC-derived IL-12 and TGF{beta}1. Our studies provide critical insight for understanding the role of the DC-based mechanical regulation of immunopathology in directing T cell lineage commitment in tumor microenvironments.

immunology↗

bioRxiv2022

Adaptive Immune Landscape of T-Cell Mediated Rejection of Human Kidney Allografts

The frequently occurring T cell mediated rejection (TCMR) is a risk factor for allograft failure. Immunosuppressive therapy fails to reverse almost 40% of TCMRs occurring in human kidney allografts. A better understanding of the molecular mechanisms of TCMR and precision therapeutics may improve allograft longevity. We investigated adaptive immune landscape of TCMR by genome wide RNA sequencing of 34 prototypic kidney allograft biopsies from 34 adult recipients of human kidney allografts. Sixteen of the 34 biopsies were categorized as Banff TCMR and the remaining 18 as Banff Normal biopsies. Computational analysis identified higher intragraft abundance of the gene sets for key players of adaptive immune system in TCMR. TCMR allografts were characterized by, i) increased antigen processing and presentation and T cell receptor signaling, ii) increased memory T cells, Tregs, Th1, Th2 and Th17 subsets, iii) increased aerobic glycolysis of lymphocytes and reduced metabolic activity of graft parenchymal cells, iv) increased T cell inhibitory receptors and exhaustion markers, v) increased apoptosis and necroptosis, and vi) increased extracellular matrix remodeling, all in comparison to Normal biopsies. Our genome-wide transcriptomics provides an atlas of adaptive immune landscape of TCMR in human kidney allografts, help deduce molecular mechanisms and prioritization of therapeutic targets.

immunology↗

bioRxiv2022

Open Access Repository-Scale Propagated Nearest Neighbor Suspect Spectral Library for Untargeted Metabolomics

Despite the increasing availability of tandem mass spectrometry (MS/MS) community spectral libraries for untargeted metabolomics over the past decade, the majority of acquired MS/MS spectra remain uninterpreted. To further aid in interpreting unannotated spectra, we created a nearest neighbor suspect spectral library, consisting of 87,916 annotated MS/MS spectra derived from hundreds of millions of public MS/MS spectra. Annotations were propagated based on structural relationships to reference molecules using MS/MS-based spectrum alignment. We demonstrate the broad relevance of the nearest neighbor suspect spectral library through representative examples of propagation-based annotation of acylcarnitines, bacterial and plant natural products, and drug metabolism. Our results also highlight how the library can help to better understand an Alzheimers brain phenotype. The nearest neighbor suspect spectral library is openly available through the GNPS platform to help investigators hypothesize candidate structures for unknown MS/MS spectra in untargeted metabolomics data.

bioinformatics↗

bioRxiv2022

Adversarial interspecies relationships facilitate population suppression by gene drive in spatially explicit models

Suppression gene drives are designed to bias their inheritance and increase in frequency in a population, disrupting an essential gene in the process. When the frequency is high enough, the population will be unable to reproduce above the replacement level and could be eliminated. CRISPR suppression drives based on the homing mechanism have already seen success in the laboratory, particularly in malaria mosquitoes. However, several models predict that the use of these drives in realistic populations with spatial structure may not achieve complete success. This is due to the ability of wild-type individuals to escape the drive and reach empty areas with reduced competition, allowing them to achieve high reproductive success and leading to extinction-recolonization cycles across the landscape. Here, we extend our continuous space gene drive framework to include two competing species or predator-prey species pairs. We find that in both general and mosquito-specific models, the presence of a competing species or predator can greatly facilitate drive-based suppression, even for drives with modest efficiency. However, the presence of a competing species also substantially increases the frequency of outcomes in which the drive is lost before suppression is achieved. These results are robust in models with seasonal population fluctuations that mosquito populations often experience. We also found that suppression can be somewhat more difficult if targeting a predator with strong predator-prey interactions. Our results illustrate the difficulty of predicting outcomes of interventions that could substantially affect populations of interacting species in complex ecosystems. However, our results are also potentially promising for the prospects of less powerful suppression gene drives for achieving successful elimination of target mosquito and other pest populations.

ecology↗

bioRxiv2022

Highly specialized carbohydrate metabolism capability in Bifidobacterium strain associated with intestinal barrier maturation in early preterm infants

"Leaky gut", or high intestinal barrier permeability, is common in preterm newborns. The role of microbiota in this process remains largely uncharacterized. We employed both short- and long-read sequencing of the 16S rRNA gene and metagenomes to characterize the intestinal microbiome of a longitudinal cohort of 113 preterm infants born between 240/7-326/7 weeks of gestation. Enabled by enhanced taxonomic resolution, we found significantly increased abundance of Bifidobacterium breve and a diet rich in mothers breastmilk to be associated with intestinal barrier maturation during the first week of life. We combined these factors using genome- resolved metagenomics and identified a highly specialized genetic capability of the Bifidobacterium strains to assimilate human milk oligosaccharides and host-derived glycoproteins. Our study proposed mechanistic roles of breastmilk feeding and intestinal microbial colonization in postnatal intestinal barrier maturation; these observations are critical towards advancing therapeutics to prevent and treat hyperpermeable gut- associated conditions, including necrotizing enterocolitis. IMPORTANCEDespite improvements in neonatal intensive care, necrotizing enterocolitis (NEC) remains a leading cause of morbidity and mortality. "Leaky gut", or intestinal barrier immaturity with elevated intestinal permeability, is the proximate cause of susceptibility to NEC. Early detection and intervention to prevent leaky gut in "at-risk" preterm neonates is critical to lower the risk for potentially life-threatening complications like NEC. However, the complex interactions between the developing gut microbial community, nutrition, and intestinal barrier function, remain largely uncharacterized. In this study, we revealed the critical role of sufficient breastmilk feeding volume and specialized carbohydrate metabolism capability of Bifidobacterium in coordinated postnatal improvement of intestinal barrier. Determining the clinical and microbial biomarkers that drive the intestinal developmental disparity will inform early detection and novel therapeutic strategies to promote appropriate intestinal barrier maturation, prevent NEC and other adverse health conditions in preterm infants.

microbiology↗

bioRxiv2022

A human embryonic limb cell atlas resolved in space and time

Human limbs emerge during the fourth post-conception week as mesenchymal buds which develop into fully-formed limbs over the subsequent months. Limb development is orchestrated by numerous temporally and spatially restricted gene expression programmes, making congenital alterations in phenotype common. Decades of work with model organisms has outlined the fundamental processes underlying vertebrate limb development, but an in-depth characterisation of this process in humans has yet to be performed. Here we detail the development of the human embryonic limb across space and time, using both single-cell and spatial transcriptomics. We demonstrate extensive diversification of cells, progressing from a restricted number of multipotent progenitors to myriad mature cell states, and identify several novel cell populations, including neural fibroblasts and multiple distinct mesenchymal states. We uncover two waves of human muscle development, each characterised by different cell states regulated by separate gene expression programmes. We identify musculin (MSC) as a key transcriptional repressor maintaining muscle stem cell identity and validate this by performing MSC knock down in human embryonic myoblasts, which results in significant upregulation of late myogenic genes. Through integration of multiple anatomically continuous spatial transcriptomic samples, we spatially map single-cell clusters across a sagittal section of a whole fetal hindlimb. We reveal a clear anatomical segregation between genes linked to brachydactyly and polysyndactyly, and uncover transcriptionally and spatially distinct populations of mesenchyme in the autopod. Finally, we perform scRNA-seq on murine embryonic limbs to facilitate cross-species developmental comparison at single-cell resolution, finding substantial homology between the two species.

developmental biology↗

bioRxiv2022

A cocktail containing two synergetic antibodies broadly neutralizes SARS-CoV-2 and its variants including Omicron BA.1 and BA.2

Neutralizing antibodies (NAbs) can prevent and treat infections caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). However, continuously emerging variants, such as Omicron, have significantly reduced the potency of most known NAbs. The selection of NAbs with broad neutralizing activities and the identification of conserved critical epitopes are still urgently needed. Here, we identified an extremely potent antibody (55A8) by single B-cell sorting from convalescent SARS-CoV-2-infected patients that recognized the receptor-binding domain (RBD) in the SARS-CoV-2 spike (S) protein. 55A8 could bind to wild-type SARS-CoV-2, Omicron BA.1 and Omicron BA.2 simultaneously with 58G6, a NAb previously identified by our group. Importantly, an antibody cocktail containing 55A8 and 58G6 (2-cocktail) showed synergetic neutralizing activity with a half-maximal inhibitory concentration (IC50) in the picomolar range in vitro and prophylactic efficacy in hamsters challenged with Omicron (BA.1) through intranasal delivery at an extraordinarily low dosage (25 g of each antibody daily) at 3 days post-infection. Structural analysis by cryo-electron microscopy (cryo-EM) revealed that 55A8 is a Class III NAb that recognizes a highly conserved epitope. It could block angiotensin-converting enzyme 2 (ACE2) binding to the RBD in the S protein trimer via steric hindrance. The epitopes in the RBD recognized by 55A8 and 58G6 were found to be different and complementary, which could explain the synergetic mechanism of these two NAbs. Our findings not only provide a potential antibody cocktail for clinical use against infection with current SARS-CoV-2 strains and future variants but also identify critical epitope information for the development of better antiviral agents.

immunology↗

bioRxiv2022

DrugRepo: A novel approach to repurpose a huge collection of compounds based on chemical and genomic features

The drug development process consumes 9-12 years and approximately one billion US dollars in terms of costs. Due to high finances and time costs required by the traditional drug discovery paradigm, repurposing the old drugs to treat cancer and rare diseases is becoming popular. Computational approaches are mainly data-driven and involve a systematic analysis of different data types leading to the formulation of repurposing hypotheses. This study presents a novel scoring algorithm based on chemical and genomic data types to repurpose vast collection of compounds for 674 cancer types and other diseases. The data types used to design the scoring algorithm are chemical structures, drug-target interactions (DTI), pathways, and disease-gene associations. The repurpose scoring algorithm is strengthened by integrating the most comprehensive manually curated datasets for each data type. More than 100 of our repurposed compounds can be matched with ongoing studies at clinical trials (https://clinicaltrials.gov/). Our analysis is supported by a web tool available at: http://drugrepo.org/.

bioinformatics↗

bioRxiv2022

Knowledge-graph-based cell-cell communication inference for spatially resolved transcriptomic data with SpaTalk

Spatially resolved transcriptomics (ST) provides genetic information in space toward elucidation of the spatial architecture in intact organs and the spatially resolved cell-cell communications mediating tissue homeostasis, development, and disease. To facilitate inference of spatially resolved cell-cell communications from ST data, we here present SpaTalk, which relies on a graph network and knowledge graph to model and score the ligand-receptor-target signaling network between spatially proximal cells, decomposed from ST data through a non-negative linear model and spatial mapping between single-cell RNA-sequencing and ST data. The performance of SpaTalk benchmarked on public single-cell ST datasets was superior to that of existing cell-cell communication inference methods. SpaTalk was then applied to STARmap, Slide-seq, and 10X Visium data, revealing the in-depth communicative mechanisms underlying normal and disease tissues with spatial structure. SpaTalk can uncover spatially resolved cell-cell communications for single-cell and spot-based ST data universally, providing new insights into spatial inter-cellular dynamics.

bioinformatics↗

bioRxiv2022

Protein 3D structure-based neural networks highly improve the accuracy in compound-protein binding affinity prediction

Due to the lack of a method to efficiently represent the multimodal information of a protein, including its structure and sequence information, predicting compound-protein binding affinity (CPA) still suffers from low accuracy when applying machine learning methods. To overcome this limitation, in a novel end-to-end architecture (named FeatNN), we develop a coevolutionary strategy to jointly represent the structure and sequence features of proteins and ultimately optimize the mathematical models for predicting CPA. Furthermore, from the perspective of data-driven approach, we proposed a rational method that can utilize both high- and low-quality databases to optimize the accuracy and generalization ability of FeatNN in CPA prediction tasks. Notably, we visually interpret the feature interaction process between sequence and structure in the rationally designed architecture. As a result, FeatNN considerably outperforms the state-of-the-art (SOTA) baseline in virtual drug screening tasks, indicating the feasibility of this approach for practical use. FeatNN provides an outstanding method for higher CPA prediction accuracy and better generalization ability by efficiently representing multimodal information of proteins via a coevolutionary strategy.

bioinformatics↗