A gene prioritization method based on a swine multi-omics knowledgebase and a deep learning model

Tabular deep learning: a comparative study applied to multi-task genome-wide prediction

PURPOSE

More accurate prediction of phenotype traits can increase the success of genomic selection in both plant and animal breeding studies and provide more reliable disease risk prediction in humans. Traditional approaches typically use regression models based on linear assumptions between the genetic markers and the traits of interest. Non-linear models have been considered as an alternative tool for modeling genomic interactions (i.e. non-additive effects) and other subtle non-linear patterns between markers and phenotype. Deep learning has become a state-of-the-art non-linear prediction method for sound, image and language data. However, genomic data is better represented in a tabular format. The existing literature on deep learning for tabular data proposes a wide range of novel architectures and reports successful results on various datasets. Tabular deep learning applications in genome-wide prediction (GWP) are still rare. In this work, we perform an overview of the main families of recent deep learning architectures for tabular data and apply them to multi-trait regression and multi-class classification for GWP on real gene datasets.

METHODS

The study involves an extensive overview of recent deep learning architectures for tabular data learning: NODE, TabNet, TabR, TabTransformer, FT-Transformer, AutoInt, GANDALF, SAINT and LassoNet. These architectures are applied to multi-trait GWP. Comprehensive benchmarks of various tabular deep learning methods are conducted to identify best practices and determine their effectiveness compared to traditional methods.

RESULTS

Extensive experimental results on several genomic datasets (three for multi-trait regression and two for multi-class classification) highlight LassoNet as a standout performer, surpassing both other tabular deep learning models and the highly efficient tree based LightGBM method in terms of both best prediction accuracy and computing efficiency.

CONCLUSION

Through series of evaluations on real-world genomic datasets, the study identifies LassoNet as a standout performer, surpassing decision tree methods like LightGBM and other tabular deep learning architectures in terms of both predictive accuracy and computing efficiency. Moreover, the inherent variable selection property of LassoNet provides a systematic way to find important genetic markers that contribute to phenotype expression.

Artificial intelligence and porcine breeding

Livestock management is evolving into a new era, characterized by the analysis of vast quantities of data (Big Data) collected from both traditional breeding methods and new technologies such as sensors, automated monitoring system, and advanced analytics. Artificial intelligence (A-In), which refers to the capability of machines to mimic human intelligence, including subfields like machine learning and deep learning, is playing a pivotal role in this transformation. A wide array of A-In techniques, successfully employed in various industrial and scientific contexts, are now being integrated into mainstream livestock management practices. In the case of swine breeding, while traditional methods have yielded considerable success, the increasing amount of information requires the adoption of new technologies such as A-In to drive productivity, enhance animal welfare, and reduce environmental impact. Current findings suggest that these techniques have the potential to match or exceed the performance of traditional methods, often being more scalable in terms of efficiency and sustainability within the breeding industry. This review provides insights into the application of A-In in porcine breeding, from the perspectives of both sows (including welfare and reproductive management) and boars (including semen quality and health), and explores new approaches which are already being applied in other species.

Machine learning methods for genomic prediction of cow behavioral traits measured by automatic milking systems in North American Holstein cattle

Identifying genome-enabled methods that provide more accurate genomic prediction is crucial when evaluating complex traits such as dairy cow behavior. In this study, we aimed to compare the predictive performance of traditional genomic prediction methods and deep learning algorithms for genomic prediction of milking refusals (MREF) and milking failures (MFAIL) in North American Holstein cows measured by automatic milking systems (milking robots). A total of 1,993,509 daily records from 4,511 genotyped Holstein cows were collected by 36 milking robot stations. After quality control, 57,600 SNPs were available for the analyses. Four genomic prediction methods were considered: Bayesian least absolute shrinkage and selection operator (LASSO), multiple layer perceptron (MLP), convolutional neural network (CNN), and GBLUP. We implemented the first 3 methods using the Keras and TensorFlow libraries in Python (v.3.9) but the GBLUP method was implemented using the BLUPF90+ family programs. The accuracy of genomic prediction (mean square error) for MREF and MFAIL was 0.34 (0.08) and 0.27 (0.08) based on LASSO, 0.36 (0.09) and 0.32 (0.09) for MLP, 0.37 (0.08) and 0.30 (0.09) for CNN, and 0.35 (0.09) and 0.31(0.09) based on GBLUP, respectively. Additionally, we observed a lower reranking of top selected individuals based on the MLP versus CNN methods compared with the other approaches for both MREF and MFAIL. Although the deep learning methods showed slightly higher accuracies than GBLUP, the results may not be sufficient to justify their use over traditional methods due to their higher computational demand and the difficulty of performing genomic prediction for nongenotyped individuals using deep learning procedures. Overall, this study provides insights into the potential feasibility of using deep learning methods to enhance genomic prediction accuracy for behavioral traits in livestock. Further research is needed to determine their practical applicability to large dairy cattle breeding programs.

Pathformer: a biological pathway informed transformer for disease diagnosis and prognosis using multi-omics data

MOTIVATION

Multi-omics data provide a comprehensive view of gene regulation at multiple levels, which is helpful in achieving accurate diagnosis of complex diseases like cancer. However, conventional integration methods rarely utilize prior biological knowledge and lack interpretability.

RESULTS

To integrate various multi-omics data of tissue and liquid biopsies for disease diagnosis and prognosis, we developed a biological pathway informed Transformer, Pathformer. It embeds multi-omics input with a compacted multi-modal vector and a pathway-based sparse neural network. Pathformer also leverages criss-cross attention mechanism to capture the crosstalk between different pathways and modalities. We first benchmarked Pathformer with 18 comparable methods on multiple cancer datasets, where Pathformer outperformed all the other methods, with an average improvement of 6.3%-14.7% in F1 score for cancer survival prediction, 5.1%-12% for cancer stage prediction, and 8.1%-13.6% for cancer drug response prediction. Subsequently, for cancer prognosis prediction based on tissue multi-omics data, we used a case study to demonstrate the biological interpretability of Pathformer by identifying key pathways and their biological crosstalk. Then, for cancer early diagnosis based on liquid biopsy data, we used plasma and platelet datasets to demonstrate Pathformer's potential of clinical applications in cancer screening. Moreover, we revealed deregulation of interesting pathways (e.g. scavenger receptor pathway) and their crosstalk in cancer patients' blood, providing potential candidate targets for cancer microenvironment study.

AVAILABILITY AND IMPLEMENTATION

Pathformer is implemented and freely available at https://github.com/lulab/Pathformer.

Identification of different myofiber types in pigs muscles and construction of regulatory networks

BACKGROUND

Skeletal muscle is composed of muscle fibers with different physiological characteristics, which plays an important role in regulating skeletal muscle metabolism, movement and body homeostasis. The type of skeletal muscle fiber directly affects meat quality. However, the transcriptome and gene interactions between different types of muscle fibers are not well understood.

RESULTS

In this paper, we selected 180-days-old Large White pigs and found that longissimus dorsi (LD) muscle was dominated by fast-fermenting myofibrils and soleus (SOL) muscle was dominated by slow-oxidizing myofibrils by frozen sections and related mRNA and protein assays. Here, we selected LD muscle and SOL muscle for transcriptomic sequencing, and identified 312 differentially expressed mRNA (DEmRs), 30 differentially expressed miRNA (DEmiRs), 183 differentially expressed lncRNA (DElRs), and 3417 differentially expressed circRNA (DEcRs). The ceRNA network included ssc-miR-378, ssc-miR-378b-3p, ssc-miR-24-3p, XR_308817, XR_308823, SMIM8, MAVS and FOS as multiple core nodes that play important roles in muscle development. Moreover, we found that different members of the miR-10 family expressed differently in oxidized and glycolytic muscle fibers, among which miR-10a-5p was highly expressed in glycolytic muscle fibers (LD) and could target MYBPH gene mRNA. Therefore, we speculate that miR-10a-5p may be involved in the transformation of muscle fiber types by targeting the MYHBP gene. In addition, PPI analysis of differentially expressed mRNA genes showed that ACTC1, ACTG2 and ACTN2 gene had the highest node degree, suggesting that this gene may play a key role in the regulatory network of muscle fiber type determination.

CONCLUSIONS

We can conclude that these genes play a key role in regulating muscle fiber type transformation. Our study provides transcriptomic profiles and ceRNA interaction networks for different muscle fiber types in pigs, providing reference for the transformation of pig muscle fiber types and the improvement of meat quality.

MOINER: A Novel Multiomics Early Integration Framework for Biomedical Classification and Biomarker Discovery

In the context of precision medicine, multiomics data integration provides a comprehensive understanding of underlying biological processes and is critical for disease diagnosis and biomarker discovery. One commonly used integration method is early integration through concatenation of multiple dimensionally reduced omics matrices due to its simplicity and ease of implementation. However, this approach is seriously limited by information loss and lack of latent feature interaction. Herein, a novel multiomics early integration framework (MOINER) based on information enhancement and image representation learning is thus presented to address the challenges. MOINER employs the self-attention mechanism to capture the intrinsic correlations of omics-features, which make it significantly outperform the existing state-of-the-art methods for multiomics data integration. Moreover, visualizing the attention embedding and identifying potential biomarkers offer interpretable insights into the prediction results. All source codes and model for MOINER are freely available https://github.com/idrblab/MOINER.

Genetic introgression from commercial European pigs to the indigenous Chinese Lijiang breed and associated changes in phenotypes

BACKGROUND

Gene flow is crucial for enhancing economic traits of livestock. In China, breeders have used hybridization strategies for decades to improve livestock performance. Here, we performed whole-genome sequencing of a native Chinese Lijiang pig (LJP) breed. By integrating previously published data, we explored the genetic structure and introgression of genetic components from commercial European pigs (EP) into the LJP, and examined the impact of this introgression on phenotypic traits.

RESULTS

Our analysis revealed significant introgression of EP breeds into the LJP and other domestic pig breeds in China. Using a haplotype-based approach, we quantified introgression levels and compared EP to LJP and other Chinese domestic pigs. The results show that EP introgression is widely prevalent in Chinese domestic pigs, although there are significant differences between breeds. We propose that LJP could potentially act as a mediator for the transmission of EP haplotypes. We also examined the correlation between EP introgression and the number of thoracic vertebrae in LJP and identified VRTN and STUM as candidate genes for this trait.

CONCLUSIONS

Our study provides evidence of introgressed European haplotypes in the LJP breed and describes the potential role of EP introgression on phenotypic changes of this indigenous breed.

Differential Regulation of Male-Hormones-Related Enhancers Revealed by Chromatin Accessibility and Transcriptional Profiles in Pig Liver

Surgical castration can effectively avoid boar taint and improve pork quality by removing the synthesis of androstenone in the testis, thereby reducing its deposition in adipose tissue. The expression of genes involved in testis-derived hormone metabolism was altered following surgical castration, but the upstream regulatory factors and underlying mechanism remain unclear. In this study, we systematically profiled chromatin accessibility and transcriptional dynamics in liver tissue of castrated and intact full-sibling Yorkshire pigs. First, we identified 897 differentially expressed genes and 6864 differential accessible regions (DARs) using RNA- and ATAC-seq. By integrating the RNA- and ATAC-seq results, 227 genes were identified, and a significant positive correlation was revealed between differential gene expression and the ATAC-seq signal. We constructed a transcription factor regulatory network after motif analysis of DARs and identified a candidate transcription factor (TF) SP1 that targeted the HSD3B1 gene, which was responsible for the metabolism of androstenone. Subsequently, we annotated DARs by incorporating H3K27ac ChIP-seq data, marking 2234 typical enhancers and 245 super enhancers involved in the regulation of all testis-derived hormones. Among these, four typical enhancers associated with HSD3B1 were identified. Furthermore, an in-depth investigation was conducted on the androstenone-related enhancers, and an androstenone-related mutation was identified in a newfound candidatetypical enhancer (andEN) with dual-luciferase assays. These findings provide further insights into how enhancers function as links between phenotypic and non-coding area variations. The discovery of upstream TF and enhancers of HSD3B1 contributes to understanding the regulatory networks of androstenone metabolism and provides an important foundation for improving pork quality.

Omics-based construction of regulatory variants can be applied to help decipher pig liver-related traits

Genetic variants can influence complex traits by altering gene expression through changes to regulatory elements. However, the genetic variants that affect the activity of regulatory elements in pigs are largely unknown, and the extent to which these variants influence gene expression and contribute to the understanding of complex phenotypes remains unclear. Here, we annotate 90,991 high-quality regulatory elements using acetylation of histone H3 on lysine 27 (H3K27ac) ChIP-seq of 292 pig livers. Combined with genome resequencing and RNA-seq data, we identify 28,425 H3K27ac quantitative trait loci (acQTLs) and 12,250 expression quantitative trait loci (eQTLs). Through the allelic imbalance analysis, we validate two causative acQTL variants in independent datasets. We observe substantial sharing of genetic controls between gene expression and H3K27ac, particularly within promoters. We infer that 46% of H3K27ac exhibit a concomitant rather than causative relationship with gene expression. By integrating GWAS, eQTLs, acQTLs, and transcription factor binding prediction, we further demonstrate their application, through metabolites dulcitol, phosphatidylcholine (PC) (16:0/16:0) and published phenotypes, in identifying likely causal variants and genes, and discovering sub-threshold GWAS loci. We provide insight into the relationship between regulatory elements and gene expression, and the genetic foundation for dissecting the molecular mechanism of phenotypes.

Haplotype-resolved 3D chromatin architecture of the hybrid pig

In diploid mammals, allele-specific three-dimensional (3D) genome architecture may lead to imbalanced gene expression. Through ultradeep in situ Hi-C sequencing of three representative somatic tissues (liver, skeletal muscle, and brain) from hybrid pigs generated by reciprocal crosses of phenotypically and physiologically divergent Berkshire and Tibetan pigs, we uncover extensive chromatin reorganization between homologous chromosomes across multiple scales. Haplotype-based interrogation of multi-omic data revealed the tissue dependence of 3D chromatin conformation, suggesting that parent-of-origin-specific conformation may drive gene imprinting. We quantify the effects of genetic variations and histone modifications on allelic differences of long-range promoter-enhancer contacts, which likely contribute to the phenotypic differences between the parental pig breeds. We also observe the fine structure of somatically paired homologous chromosomes in the pig genome, which has a functional implication genome-wide. This work illustrates how allele-specific chromatin architecture facilitates concomitant shifts in allele-biased gene expression, as well as the possible consequential phenotypic changes in mammals.

Veterinary systems biology for bridging the phenotype-genotype gap via computational modeling for disease epidemiology and animal welfare

Veterinary systems biology is an innovative approach that integrates biological data at the molecular and cellular levels, allowing for a more extensive understanding of the interactions and functions of complex biological systems in livestock and veterinary science. It has tremendous potential to integrate multi-omics data with the support of vetinformatics resources for bridging the phenotype-genotype gap via computational modeling. To understand the dynamic behaviors of complex systems, computational models are frequently used. It facilitates a comprehensive understanding of how a host system defends itself against a pathogen attack or operates when the pathogen compromises the host's immune system. In this context, various approaches, such as systems immunology, network pharmacology, vaccinology and immunoinformatics, can be employed to effectively investigate vaccines and drugs. By utilizing this approach, we can ensure the health of livestock. This is beneficial not only for animal welfare but also for human health and environmental well-being. Therefore, the current review offers a detailed summary of systems biology advancements utilized in veterinary sciences, demonstrating the potential of the holistic approach in disease epidemiology, animal welfare and productivity.

AGIDB: a versatile database for genotype imputation and variant decoding across species

The high cost of large-scale, high-coverage whole-genome sequencing has limited its application in genomics and genetics research. The common approach has been to impute whole-genome sequence variants obtained from a few individuals for a larger population of interest individually genotyped using SNP chip. An alternative involves low-coverage whole-genome sequencing (lcWGS) of all individuals in the larger population, followed by imputation to sequence resolution. To overcome limitations of processing lcWGS data and meeting specific genotype imputation requirements, we developed AGIDB (https://agidb.pro), a website comprising tools and database with an unprecedented sample size and comprehensive variant decoding for animals. AGIDB integrates whole-genome sequencing and chip data from 17 360 and 174 945 individuals, respectively, across 89 species to identify over one billion variants, totaling a massive 688.57 TB of processed data. AGIDB focuses on integrating multiple genotype imputation scenarios. It also provides user-friendly searching and data analysis modules that enable comprehensive annotation of genetic variants for specific populations. To meet a wide range of research requirements, AGIDB offers downloadable reference panels for each species in addition to its extensive dataset, variant decoding and utility tools. We hope that AGIDB will become a key foundational resource in genetics and breeding, providing robust support to researchers.

Searching for new signals for susceptibility to umbilical hernia through genome-wide association analysis in three pig breeds

Umbilical hernia (UH) is a prevalent congenital disorder in pigs, resulting in considerable economic losses and severe animal welfare issues. In the present study, we conducted a genome-wide association study (GWAS) using the GeneSeek 50K Chip in 2777 pigs (Duroc, n = 1267; Landrace, n = 696; and Yorkshire, n = 814) to explore the candidate genes underlying the risk of umbilical hernia in pigs. After quality control analyses, 2748 animals and 48 524 single nucleotide polymorphisms (SNPs) were retained for subsequent GWAS analysis using the FarmCPU model. The heritability of umbilical hernias was estimated to 0.51 ± 0.04, indicating a reasonable basis for investigating genetic markers associated with this disorder. We identified 54 SNPs and 517 candidate genes that showed significant associations with susceptibility to umbilical hernia across the combined population of the three pig breeds. Gene enrichment analyses highlighted several crucial pathways for platelet degranulation, inflammatory mediator regulation of TRP channels and ion transport. These findings provide further insights into the underlying genetic architecture of umbilical hernias in pigs.

Polygenic Risk Score for Cardiovascular Diseases in Artificial Intelligence Paradigm: A Review

Cardiovascular disease (CVD) related mortality and morbidity heavily strain society. The relationship between external risk factors and our genetics have not been well established. It is widely acknowledged that environmental influence and individual behaviours play a significant role in CVD vulnerability, leading to the development of polygenic risk scores (PRS). We employed the PRISMA search method to locate pertinent research and literature to extensively review artificial intelligence (AI)-based PRS models for CVD risk prediction. Furthermore, we analyzed and compared conventional vs. AI-based solutions for PRS. We summarized the recent advances in our understanding of the use of AI-based PRS for risk prediction of CVD. Our study proposes three hypotheses: i) Multiple genetic variations and risk factors can be incorporated into AI-based PRS to improve the accuracy of CVD risk predicting. ii) AI-based PRS for CVD circumvents the drawbacks of conventional PRS calculators by incorporating a larger variety of genetic and non-genetic components, allowing for more precise and individualised risk estimations. iii) Using AI approaches, it is possible to significantly reduce the dimensionality of huge genomic datasets, resulting in more accurate and effective disease risk prediction models. Our study highlighted that the AI-PRS model outperformed traditional PRS calculators in predicting CVD risk. Furthermore, using AI-based methods to calculate PRS may increase the precision of risk predictions for CVD and have significant ramifications for individualized prevention and treatment plans.

Young SINEs in pig genomes impact gene regulation, genetic diversity, and complex traits

Transposable elements (TEs) are a major source of genetic polymorphisms and play a role in chromatin architecture, gene regulatory networks, and genomic evolution. However, their functional role in pigs and contributions to complex traits are largely unknown. We created a catalog of TEs (n = 3,087,929) in pigs and found that young SINEs were predominantly silenced by histone modifications, DNA methylation, and decreased accessibility. However, some transcripts from active young SINEs showed high tissue-specificity, as confirmed by analyzing 3570 RNA-seq samples. We also detected 211,067 dimorphic SINEs in 374 individuals, including 340 population-specific ones associated with local adaptation. Mapping these dimorphic SINEs to genome-wide associations of 97 complex traits in pigs, we found 54 candidate genes (e.g., ANK2 and VRTN) that might be mediated by TEs. Our findings highlight the important roles of young SINEs and provide a supplement for genotype-to-phenotype associations and modern breeding in pigs.

E-GWAS: an ensemble-like GWAS strategy that provides effective control over false positive rates without decreasing true positives

BACKGROUND

Genome-wide association studies (GWAS) are an effective way to explore genotype-phenotype associations in humans, animals, and plants. Various GWAS methods have been developed based on different genetic or statistical assumptions. However, no single method is optimal for all traits and, for many traits, the putative single nucleotide polymorphisms (SNPs) that are detected by the different methods do not entirely overlap due to the diversity of the genetic architecture of complex traits. Therefore, multi-tool-based GWAS strategies that combine different methods have been increasingly employed. To take this one step further, we propose an ensemble-like GWAS strategy (E-GWAS) that statistically integrates GWAS results from different single GWAS methods.

RESULTS

E-GWAS was compared with various single GWAS methods using simulated phenotype traits with different genetic architectures. E-GWAS performed stably across traits with different genetic architectures and effectively controlled the number of false positive genetic variants detected without decreasing the number of true positive variants. In addition, its performance could be further improved by using a bin-merged strategy and the addition of more distinct single GWAS methods. Our results show that the numbers of true and false positive SNPs detected by the E-GWAS strategy slightly increased and decreased, respectively, with increasing bin size and when the number and the diversity of individual GWAS methods that were integrated in E-GWAS increased, the latter being more effective than the bin-merged strategy. The E-GWAS strategy was also applied to a real dataset to study backfat thickness in a pig population, and 10 candidate genes related to this trait and expressed in adipose-associated tissues were identified.

CONCLUSIONS

Using both simulated and real datasets, we show that E-GWAS is a reliable and robust strategy that effectively integrates the GWAS results of different methods and reduces the number of false positive SNPs without decreasing that of true positive SNPs.

Rise of Deep Learning Clinical Applications and Challenges in Omics Data: A Systematic Review

This research aims to review and evaluate the most relevant scientific studies about deep learning (DL) models in the omics field. It also aims to realize the potential of DL techniques in omics data analysis fully by demonstrating this potential and identifying the key challenges that must be addressed. Numerous elements are essential for comprehending numerous studies by surveying the existing literature. For example, the clinical applications and datasets from the literature are essential elements. The published literature highlights the difficulties encountered by other researchers. In addition to looking for other studies, such as guidelines, comparative studies, and review papers, a systematic approach is used to search all relevant publications on omics and DL using different keyword variants. From 2018 to 2022, the search procedure was conducted on four Internet search engines: IEEE Xplore, Web of Science, ScienceDirect, and PubMed. These indexes were chosen because they offer enough coverage and linkages to numerous papers in the biological field. A total of 65 articles were added to the final list. The inclusion and exclusion criteria were specified. Of the 65 publications, 42 are clinical applications of DL in omics data. Furthermore, 16 out of 65 articles comprised the review publications based on single- and multi-omics data from the proposed taxonomy. Finally, only a small number of articles (7/65) were included in papers focusing on comparative analysis and guidelines. The use of DL in studying omics data presented several obstacles related to DL itself, preprocessing procedures, datasets, model validation, and testbed applications. Numerous relevant investigations were performed to address these issues. Unlike other review papers, our study distinctly reflects different observations on omics with DL model areas. We believe that the result of this study can be a useful guideline for practitioners who look for a comprehensive view of the role of DL in omics data analysis.

IAnimal: a cross-species omics knowledgebase for animals

With the exponential growth of multi-omics data, its integration and utilization have brought unprecedented opportunities for the interpretation of gene regulation mechanisms and the comprehensive analyses of biological systems. IAnimal (https://ianimal.pro/), a cross-species, multi-omics knowledgebase, was developed to improve the utilization of massive public data and simplify the integration of multi-omics information to mine the genetic mechanisms of objective traits. Currently, IAnimal provides 61 191 individual omics data of genome (WGS), transcriptome (RNA-Seq), epigenome (ChIP-Seq, ATAC-Seq) and genome annotation information for 21 species, such as mice, pigs, cattle, chickens, and macaques. The scale of its total clean data has reached 846.46 TB. To better understand the biological significance of omics information, a deep learning model for IAnimal was built based on BioBERT and AutoNER to mine 'gene' and 'trait' entities from 2 794 237 abstracts, which has practical significance for comprehending how each omics layer regulates genes to affect traits. By means of user-friendly web interfaces, flexible data application programming interfaces, and abundant functional modules, IAnimal enables users to easily query, mine, and visualize characteristics in various omics, and to infer how genes play biological roles under the influence of various omics layers.

Animal-SNPAtlas: a comprehensive SNP database for multiple animals

Single-nucleotide polymorphisms (SNPs) as the most important type of genetic variation are widely used in describing population characteristics and play vital roles in animal genetics and breeding. Large amounts of population genetic variation resources and tools have been developed in human, which provided solid support for human genetic studies. However, compared with human, the development of animal genetic variation databases was relatively slow, which limits the genetic researches in these animals. To fill this gap, we systematically identified ∼ 499 million high-quality SNPs from 4784 samples of 20 types of animals. On that basis, we annotated the functions of SNPs, constructed high-density reference panels and calculated genome-wide linkage disequilibrium (LD) matrixes. We further developed Animal-SNPAtlas, a user-friendly database (http://gong_lab.hzau.edu.cn/Animal_SNPAtlas/) which includes high-quality SNP datasets and several support tools for multiple animals. In Animal-SNPAtlas, users can search the functional annotation of SNPs, perform online genotype imputation, explore and visualize LD information, browse variant information using the genome browser and download SNP datasets for each species. With the massive SNP datasets and useful tools, Animal-SNPAtlas will be an important fundamental resource for the animal genomics, genetics and breeding community.

Feature importance network reveals novel functional relationships between biological features in Arabidopsis thaliana

Understanding how the different cellular components are working together to form a living cell requires multidisciplinary approaches combining molecular and computational biology. Machine learning shows great potential in life sciences, as it can find novel relationships between biological features. Here, we constructed a dataset of 11,801 gene features for 31,522 Arabidopsis thaliana genes and developed a machine learning workflow to identify linked features. The detected linked features are visualised as a Feature Important Network (FIN), which can be mined to reveal a variety of novel biological insights pertaining to gene function. We demonstrate how FIN can be used to generate novel insights into gene function. To make this network easily accessible to the scientific community, we present the FINder database, available at finder.plant.tools.

Integration of multi-omics data reveals cis-regulatory variants that are associated with phenotypic differentiation of eastern from western pigs

Background

The genetic mechanisms that underlie phenotypic differentiation in breeding animals have important implications in evolutionary biology and agriculture. However, the contribution of cis-regulatory variants to pig phenotypes is poorly understood. Therefore, our aim was to elucidate the molecular mechanisms by which non-coding variants cause phenotypic differences in pigs by combining evolutionary biology analyses and functional genomics.

Results

We obtained a high-resolution phased chromosome-scale reference genome with a contig N50 of 18.03 Mb for the Luchuan pig breed (a representative eastern breed) and profiled potential selective sweeps in eastern and western pigs by resequencing the genomes of 234 pigs. Multi-tissue transcriptome and chromatin accessibility analyses of these regions suggest that tissue-specific selection pressure is mediated by promoters and distal cis-regulatory elements. Promoter variants that are associated with increased expression of the lysozyme (LYZ) gene in the small intestine might enhance the immunity of the gastrointestinal tract and roughage tolerance in pigs. In skeletal muscle, an enhancer-modulating single-nucleotide polymorphism that is associated with up-regulation of the expression of the troponin C1, slow skeletal and cardiac type (TNNC1) gene might increase the proportion of slow muscle fibers and affect meat quality.

Conclusions

Our work sheds light on the molecular mechanisms by which non-coding variants shape phenotypic differences in pigs and provides valuable resources and novel perspectives to dissect the role of gene regulatory evolution in animal domestication and breeding.

A benchmark study of deep learning-based multi-omics data fusion methods for cancer

BACKGROUND

A fused method using a combination of multi-omics data enables a comprehensive study of complex biological processes and highlights the interrelationship of relevant biomolecules and their functions. Driven by high-throughput sequencing technologies, several promising deep learning methods have been proposed for fusing multi-omics data generated from a large number of samples.

RESULTS

In this study, 16 representative deep learning methods are comprehensively evaluated on simulated, single-cell, and cancer multi-omics datasets. For each of the datasets, two tasks are designed: classification and clustering. The classification performance is evaluated by using three benchmarking metrics including accuracy, F1 macro, and F1 weighted. Meanwhile, the clustering performance is evaluated by using four benchmarking metrics including the Jaccard index (JI), C-index, silhouette score, and Davies Bouldin score. For the cancer multi-omics datasets, the methods' strength in capturing the association of multi-omics dimensionality reduction results with survival and clinical annotations is further evaluated. The benchmarking results indicate that moGAT achieves the best classification performance. Meanwhile, efmmdVAE, efVAE, and lfmmdVAE show the most promising performance across all complementary contexts in clustering tasks.

CONCLUSIONS

Our benchmarking results not only provide a reference for biomedical researchers to choose appropriate deep learning-based multi-omics data fusion methods, but also suggest the future directions for the development of more effective multi-omics data fusion methods. The deep learning frameworks are available at https://github.com/zhenglinyi/DL-mo .

Enhancer-promoter interaction maps provide insights into skeletal muscle-related traits in pig genome

Background

Gene expression programs are intimately linked to the interplay of active cis regulatory elements mediated by chromatin contacts and associated RNAs. Genome-wide association studies (GWAS) have identified many variants in these regulatory elements that can contribute to phenotypic diversity. However, the functional interpretation of these variants remains nontrivial due to the lack of chromatin contact information or limited contact resolution. Furthermore, the distribution and role of chromatin-associated RNAs in gene expression and chromatin conformation remain poorly understood. To address this, we first present a comprehensive interaction map of nuclear dynamics of 3D chromatin-chromatin interactions (H3K27ac BL-HiChIP) and RNA-chromatin interactions (GRID-seq) to reveal genomic variants that contribute to complex skeletal muscle traits.

Results

In a genome-wide scan, we provide systematic fine mapping and gene prioritization from GWAS leading signals that underlie phenotypic variability of growth rate, meat quality, and carcass performance. A set of candidate functional variants and 54 target genes previously not detected were identified, with 71% of these candidate functional variants choosing to skip over their nearest gene to regulate the target gene in a long-range manner. The effects of three functional variants regulating KLF6 (related to days to 100 kg), MXRA8 (related to lean meat percentage), and TAF11 (related to loin muscle depth) were observed in two pig populations. Moreover, we find that this multi-omics interaction map consists of functional communities that are enriched in specific biological functions, and GWAS target genes can serve as core genes for exploring peripheral trait-relevant genes.

Conclusions

Our results provide a valuable resource of candidate functional variants for complex skeletal muscle-related traits and establish an integrated approach to complement existing 3D genomics by exploiting RNA-chromatin and chromatin-chromatin interactions for future association studies.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12915-022-01322-2.

Multimodal deep learning for biomedical data fusion: a review

Biomedical data are becoming increasingly multimodal and thereby capture the underlying complex relationships among biological processes. Deep learning (DL)-based data fusion strategies are a popular approach for modeling these nonlinear relationships. Therefore, we review the current state-of-the-art of such methods and propose a detailed taxonomy that facilitates more informed choices of fusion strategies for biomedical applications, as well as research on novel methods. By doing so, we find that deep fusion strategies often outperform unimodal and shallow approaches. Additionally, the proposed subcategories of fusion strategies show different advantages and drawbacks. The review of current methods has shown that, especially for intermediate fusion strategies, joint representation learning is the preferred approach as it effectively models the complex interactions of different levels of biological organization. Finally, we note that gradual fusion, based on prior biological knowledge or on search strategies, is a promising future research path. Similarly, utilizing transfer learning might overcome sample size limitations of multimodal data sets. As these data sets become increasingly available, multimodal DL approaches present the opportunity to train holistic models that can learn the complex regulatory dynamics behind health and disease.

Rapid Visual CRISPR Assay: A Naked-Eye Colorimetric Detection Method for Nucleic Acids Based on CRISPR/Cas12a and a Convolutional Neural Network

Rapid diagnosis based on naked-eye colorimetric detection remains challenging, but it could build new capacities for molecular point-of-care testing (POCT). In this study, we evaluated the performance of 16 types of single-stranded DNA-fluorophore-quencher (ssDNA-FQ) reporters for use with clusters of regularly spaced short palindrome repeats (CRISPR)/Cas12a-based visual colorimetric assays. Among them, nine ssDNA-FQ reporters were found to be suitable for direct visual colorimetric detection, with especially very strong performance using ROX-labeled reporters. We optimized the reaction concentrations of these ssDNA-FQ reporters for a naked-eye read-out of assay results (no transducing component required for visualization). In particular, we developed a convolutional neural network algorithm to standardize and automate the analytical colorimetric assessment of images and integrated this into the MagicEye mobile phone software. A field-deployable assay platform named RApid VIsual CRISPR (RAVI-CRISPR) based on a ROX-labeled reporter with isothermal amplification and CRISPR/Cas12a targeting was established. We deployed RAVI-CRISPR in a single tube toward an instrument-less colorimetric POCT format that required only a portable rechargeable hand warmer for incubation. The RAVI-CRISPR was successfully used for the high-sensitivity detection of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and African swine fever virus (ASFV). Our study demonstrates this RAVI-CRISPR/MagicEye system to be suitable for distinguishing different pathogenic nucleic acid targets with high specificity and sensitivity as the simplest-to-date platform for rapid pen- or bed-side testing.

A roadmap for multi-omics data integration using deep learning

High-throughput next-generation sequencing now makes it possible to generate a vast amount of multi-omics data for various applications. These data have revolutionized biomedical research by providing a more comprehensive understanding of the biological systems and molecular mechanisms of disease development. Recently, deep learning (DL) algorithms have become one of the most promising methods in multi-omics data analysis, due to their predictive performance and capability of capturing nonlinear and hierarchical features. While integrating and translating multi-omics data into useful functional insights remain the biggest bottleneck, there is a clear trend towards incorporating multi-omics analysis in biomedical research to help explain the complex relationships between molecular layers. Multi-omics data have a role to improve prevention, early detection and prediction; monitor progression; interpret patterns and endotyping; and design personalized treatments. In this review, we outline a roadmap of multi-omics integration using DL and offer a practical perspective into the advantages, challenges and barriers to the implementation of DL in multi-omics data.

Mapping short tandem repeats for liver gene expression traits helps prioritize potential causal variants for complex traits in pigs

BACKGROUND

Short tandem repeats (STRs) were recently found to have significant impacts on gene expression and diseases in humans, but their roles on gene expression and complex traits in pigs remain unexplored. This study investigates the effects of STRs on gene expression in liver tissues based on the whole-genome sequences and RNA-Seq data of a discovery cohort of 260 F6 individuals and a validation population of 296 F7 individuals from a heterogeneous population generated from crosses among eight pig breeds.

RESULTS

We identified 5203 and 5868 significantly expression STRs (eSTRs, FDR < 1%) in the F6 and F7 populations, respectively, most of which could be reciprocally validated (π1 = 0.92). The eSTRs explained 27.5% of the cis-heritability of gene expression traits on average. We further identified 235 and 298 fine-mapped STRs through the Bayesian fine-mapping approach in the F6 and F7 pigs, respectively, which were significantly enriched in intron, ATAC peak, compartment A and H3K4me3 regions. We identified 20 fine-mapped STRs located in 100 kb windows upstream and downstream of published complex trait-associated SNPs, which colocalized with epigenetic markers such as H3K27ac and ATAC peaks. These included eSTR of the CLPB, PGLS, PSMD6 and DHDH genes, which are linked with genome-wide association study (GWAS) SNPs for blood-related traits, leg conformation, growth-related traits, and meat quality traits, respectively.

CONCLUSIONS

This study provides insights into the effects of STRs on gene expression traits. The identified eSTRs are valuable resources for prioritizing causal STRs for complex traits in pigs.

A large-scale genome-wide association analysis reveals QTL and candidate genes for intramuscular fat content in Duroc pigs

This study aimed at identifying genomic regions and genes associated with intramuscular fat content (IMF) in Duroc pigs using a weighted single-step GWAS. Data from 3912 pigs, of which 3770 animals were genotyped with GeneSeek Porcine 50K Bead chip, were used for the association analysis. We identified 19 genomic regions that each explained >1% of the additive genetic variance associated with IMF. Notably, a consistent QTL on SSC7 (117.42-117.92 Mb) was confirmed, explaining 3.70% of the additive genetic variance, and two genes, BDKRB2 and ATG2B, were highlighted as promising candidates for IMF. Two QTL (SSC7, 94.19-94.64 Mb; SSC14, 123.25-123.75 Mb), which harbored MED6 and MAP3K9 genes and TCF7L2 gene respectively, were newly identified as associated with IMF. In conclusion, we identified a consistent QTL and additional genomic regions and genes that contributed to the genetic variance of IMF using a large-scale sample size of genotyped pigs and genealogical information.

Genome-wide association and transcriptome studies identify candidate genes and pathways for feed conversion ratio in pigs

Background

The feed conversion ratio (FCR) is an important productive trait that greatly affects profits in the pig industry. Elucidating the genetic mechanisms underpinning FCR may promote more efficient improvement of FCR through artificial selection. In this study, we integrated a genome-wide association study (GWAS) with transcriptome analyses of different tissues in Yorkshire pigs (YY) with the aim of identifying key genes and signalling pathways associated with FCR.

Results

A total of 61 significant single nucleotide polymorphisms (SNPs) were detected by GWAS in YY. All of these SNPs were located on porcine chromosome (SSC) 5, and the covered region was considered a quantitative trait locus (QTL) region for FCR. Some genes distributed around these significant SNPs were considered as candidates for regulating FCR, including TPH2, FAR2, IRAK3, YARS2, GRIP1, FRS2, CNOT2 and TRHDE. According to transcriptome analyses in the hypothalamus, TPH2 exhibits the potential to regulate intestinal motility through serotonergic synapse and oxytocin signalling pathways. In addition, GRIP1 may be involved in glutamatergic and GABAergic signalling pathways, which regulate FCR by affecting appetite in pigs. Moreover, GRIP1, FRS2, CNOT2, and TRHDE may regulate metabolism in various tissues through a thyroid hormone signalling pathway.

Conclusions

Based on the results from GWAS and transcriptome analyses, the TPH2, GRIP1, FRS2, TRHDE, and CNOT2 genes were considered candidate genes for regulating FCR in Yorkshire pigs. These findings improve the understanding of the genetic mechanisms of FCR and may help optimize the design of breeding schemes.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12864-021-07570-w.

A compendium and comparative epigenomics analysis of cis-regulatory elements in the pig genome

Although major advances in genomics have initiated an exciting new era of research, a lack of information regarding cis-regulatory elements has limited the genetic improvement or manipulation of pigs as a meat source and biomedical model. Here, we systematically characterize cis-regulatory elements and their functions in 12 diverse tissues from four pig breeds by adopting similar strategies as the ENCODE and Roadmap Epigenomics projects, which include RNA-seq, ATAC-seq, and ChIP-seq. In total, we generate 199 datasets and identify more than 220,000 cis-regulatory elements in the pig genome. Surprisingly, we find higher conservation of cis-regulatory elements between human and pig genomes than those between human and mouse genomes. Furthermore, the differences of topologically associating domains between the pig and human genomes are associated with morphological evolution of the head and face. Beyond generating a major new benchmark resource for pig epigenetics, our study provides basic comparative epigenetic data relevant to using pigs as models in human biomedical research.

Improvement, identification, and target prediction for miRNAs in the porcine genome by using massive, public high-throughput sequencing data

Despite the broad variety of available microRNA (miRNA) research tools and methods, their application to the identification, annotation, and target prediction of miRNAs in nonmodel organisms is still limited. In this study, we collected nearly all public sRNA-seq data to improve the annotation for known miRNAs and identify novel miRNAs that have not been annotated in pigs (Sus scrofa). We newly annotated 210 mature sequences in known miRNAs and found that 43 of the known miRNA precursors were problematic due to redundant/missing annotations or incorrect sequences. We also predicted 811 novel miRNAs with high confidence, which was twice the current number of known miRNAs for pigs in miRBase. In addition, we proposed a correlation-based strategy to predict target genes for miRNAs by using a large amount of sRNA-seq and RNA-seq data. We found that the correlation-based strategy provided additional evidence of expression compared with traditional target prediction methods. The correlation-based strategy also identified the regulatory pairs that were controlled by nonbinding sites with a particular pattern, which provided abundant complementarity for studying the mechanism of miRNAs that regulate gene expression. In summary, our study improved the annotation of known miRNAs, identified a large number of novel miRNAs, and predicted target genes for all pig miRNAs by using massive public data. This large data-based strategy is also applicable for other nonmodel organisms with incomplete annotation information.

Positive selection signatures in Anqing six-end-white pig population based on reduced-representation genome sequencing data

Anqing six-end-white (AQ) pig performs well on resistance to coarse fodder and disease, reproduction and meat quality, offering high potential for exploitation. Environmental conditions and strict selections from local farmers have cultivated the AQ pig to be an outstanding and unique local pig breed. Thus we aim to detect genetic positive selection signatures within the AQ pig population to explore underlying genetic mechanisms. A relative extended haplotype homozygosity (REHH) test was performed in the population of 79 AQ pigs to seek evidence demonstrating that selective actions have left an imprint on the whole genome. In total, 430 500 REHH tests were performed on 53 067 core regions with average REHH tests of 8.11, average lengths of 11.50 kb and an overall length of 610.38 Mb which accounted for 26.94% of the whole genome. Finally, a total of 1819 core haplotypes (P < 0.01) and 586 candidate genes were obtained. These genes were mainly related to meat quality (MYOG, SNX19), resistance to disease (CRISPLD2, CD14) and reproduction traits (ERBB2, NRP2). A panel of genes within the 30 top significant REHH tests was mainly categorized to traits of meat quality and disease resistance. Among 13 KEGG pathways, MAPK, GnRH and Oxytocin signaling pathways, associated with the biological processes of crucial economic traits, were noteworthy. The excellent characteristics of the AQ pig benefited from the combination of natural and human factors. We provide a sketch map that shows the distribution of selection footprints on the whole genome of AQ pig and found potential genes for future studies.