On the holobiont 'predictome' of immunocompetence in pigs.

BACKGROUND: Gut microbial composition plays an important role in numerous traits, including immune response. Integration of host genomic information with microbiome data is a natural step in the prediction of complex traits, although methods to optimize this are still largely unexplored. In this paper, we assess the impact of different modelling strategies on the predictive capacity for six porcine immunocompetence traits when both genotype and microbiota data are available. METHODS: We used phenotypic data on six immunity traits and the relative abundance of gut bacterial communities on 400 Duroc pigs that were genotyped for 70 k SNPs. We compared the predictive accuracy, defined as the correlation between predicted and observed phenotypes, of a wide catalogue of models: reproducing kernel Hilbert space (RKHS), Bayes C, and an ensemble method, using a range of priors and microbial clustering strategies. Combined (holobiont) models that include both genotype and microbiome data were compared with partial models that use one source of variation only. RESULTS: Overall, holobiont models performed better than partial models. Host genotype was especially relevant for predicting adaptive immunity traits (i.e., concentration of immunoglobulins M and G), whereas microbial composition was important for predicting innate immunity traits (i.e., concentration of haptoglobin and C-reactive protein and lymphocyte phagocytic capacity). None of the models was uniformly best across all traits. We observed a greater variability in predictive accuracies across models when microbiability (the variance explained by the microbiome) was high. Clustering microbial abundances did not necessarily increase predictive accuracy. CONCLUSIONS: Gut microbiota information is useful for predicting immunocompetence traits, especially those related to innate immunity. Modelling microbiome abundances deserves special attention when microbiability is high. Clustering microbial data for prediction is not recommended by default.

Effects of extremely low-frequency magnetic fields on human MDA-MB-231 breast cancer cells: proteomic characterization.

Extremely low-frequency electromagnetic fields (ELF-MF) can modify the cell viability and regulatory processes of some cell types, including breast cancer cells. Breast cancer is a multifactorial disease where a role for ELF-MF cannot be excluded. ELF-MF may influence the biological properties of breast cells through molecular mechanisms and signaling pathways that are still unclear. This study analyzed the changes in the cell viability, cellular morphology, oxidative stress response and alteration of proteomic profile in breast cancer cells (MDA-MB-231) exposed to ELF-MF (50 Hz, 1 mT for 4 h). Non-tumorigenic human breast cells (MCF-10A) were used as control cells. Exposed MDA-MB-231 breast cancer cells increased their viability and live cell number and showed a higher density and length of filopodia compared with the unexposed cells. In addition, ELF-MF induced an increase of the mitochondrial ROS levels and an alteration of mitochondrial morphology. Proteomic data analysis showed that ELF-MF altered the expression of 328 proteins in MDA-MB-231 cells and of 242 proteins in MCF-10A cells. Gene Ontology term enrichment analysis demonstrated that in both cell lines ELF-MF exposure up-regulated the genes enriched in "focal adhesion" and "mitochondrion". The ELF-MF exposure decreased the adhesive properties of MDA-MB-231 cells and increased the migration and invasion cell abilities. At the same time, proteomic analysis, confirmed by Real Time PCR, revealed that transcription factors associated with cellular reprogramming were upregulated in MDA-MB-231 cells and downregulated in MCF-10A cells after ELF-MF exposure. MDA-MB-231 breast cancer cells exposed to 1 mT 50 Hz ELF-MF showed modifications in proteomic profile together with changes in cell viability, cellular morphology, oxidative stress response, adhesion, migration and invasion cell abilities. The main signaling pathways involved were relative to focal adhesion, mitochondrion and cellular reprogramming.

Alterations in Pregnenolone and Testosterone Levels in Male Shift Workers.

Steroid hormone levels are closely related to the endogenous circadian rhythm induced by sleep-wake and dark-light cycles. Shift work that disrupts the circadian rhythm may influence the levels of steroid hormones. The association between shift work and alterations in female sex steroid hormone levels has been studied, but little is known about testosterone and its precursor pregnenolone levels in male shift workers. The present study investigated serum pregnenolone and testosterone levels in a group of shift and daytime male workers. All participants were sampled at the beginning of the morning shift. Lower levels of serum pregnenolone and total testosterone were found in the shift workers compared to the daytime workers. Variations in pregnenolone levels may have consequences for well-being, and they might produce consequences for the levels of hormones downstream of the steroid hormone cascade, such as testosterone. The low levels of testosterone found in shift workers demonstrate the perturbative effect of shift work on testosterone serum levels, which may be independent and/or related to pregnenolone synthesis.

Modeling microRNA-driven post-transcriptional regulation using exon-intron split analysis in pigs.

The contribution of microRNAs (miRNAs) to mRNA post-transcriptional regulation has often been explored by the post hoc selection of downregulated genes and determining whether they harbor binding sites for miRNAs of interest. This approach, however, does not discriminate whether these mRNAs are also downregulated at the transcriptional level. Here, we have characterized the transcriptional and post-transcriptional changes in mRNA expression in two porcine tissues: gluteus medius muscle of fasted and fed Duroc gilts and adipose tissue of lean and obese Duroc-Göttingen minipigs. Exon-intron split analysis of RNA-seq data allowed us to identify downregulated mRNAs with high post-transcriptional signals in fed or obese states, and we assessed whether they harbor binding sites for upregulated miRNAs in any of these two physiological states. We found 26 downregulated mRNAs with high post-transcriptional signals in the muscle of fed gilts and 21 of these were predicted targets of miRNAs upregulated in fed pigs. For adipose tissue, 44 downregulated mRNAs in obese minipigs displayed high post-transcriptional signals, and 25 of these were predicted targets of miRNAs upregulated in the obese state. These results suggest that the contribution of miRNAs to mRNA repression is more prominent in the skeletal muscle system. Finally, we identified several genes that may play relevant roles in the energy homeostasis of the pig skeletal muscle (DKK2 and PDK4) and adipose (SESN3 and ESRRG) tissues. By differentiating transcriptional from post-transcriptional changes in mRNA expression, exon-intron split analysis provides a valuable view of the regulation of gene expression, complementary to canonical differential expression analyses.

Corrigendum to "Automatic Fruit Morphology Phenome and Genetic Analysis: An Application in the Octoploid Strawberry".

[This corrects the article DOI: 10.34133/2021/9812910.].

Leveraging host-genetics and gut microbiota to determine immunocompetence in pigs.

BACKGROUND: The gut microbiota influences host performance playing a relevant role in homeostasis and function of the immune system. The aim of the present work was to identify microbial signatures linked to immunity traits and to characterize the contribution of host-genome and gut microbiota to the immunocompetence in healthy pigs. RESULTS: To achieve this goal, we undertook a combination of network, mixed model and microbial-wide association studies (MWAS) for 21 immunity traits and the relative abundance of gut bacterial communities in 389 pigs genotyped for 70K SNPs. The heritability (h2; proportion of phenotypic variance explained by the host genetics) and microbiability (m2; proportion of variance explained by the microbial composition) showed similar values for most of the analyzed immunity traits, except for both IgM and IgG in plasma that was dominated by the host genetics, and the haptoglobin in serum which was the trait with larger m2 (0.275) compared to h2 (0.138). Results from the MWAS suggested a polymicrobial nature of the immunocompetence in pigs and revealed associations between pigs gut microbiota composition and 15 of the analyzed traits. The lymphocytes phagocytic capacity (quantified as mean fluorescence) and the total number of monocytes in blood were the traits associated with the largest number of taxa (6 taxa). Among the associations identified by MWAS, 30% were confirmed by an information theory network approach. The strongest confirmed associations were between Fibrobacter and phagocytic capacity of lymphocytes (r = 0.37), followed by correlations between Streptococcus and the percentage of phagocytic lymphocytes (r = -0.34) and between Megasphaera and serum concentration of haptoglobin (r = 0.26). In the interaction network, Streptococcus and percentage of phagocytic lymphocytes were the keystone bacterial and immune-trait, respectively. CONCLUSIONS: Overall, our findings reveal an important connection between gut microbiota composition and immunity traits in pigs, and highlight the need to consider both sources of information, host genome and microbial levels, to accurately characterize immunocompetence in pigs.

Influence of Work on Andropause and Menopause: A Systematic Review.

Aging is associated with gender-specific hormonal changes that progressively lead to gonadal insufficiency, a condition which characterizes a minority of men and all women. Work-related factors, such as stress and pollutant exposure, affect gonadal function and can interfere with reproduction in both genders. A systematic review of the PubMed, SCOPUS and EMBASE databases was conducted, according to the Preferred Reporting Items for Systemic Reviews and Meta-Analyses (PRISMA) statement to investigate the effect of occupational factors on andropause and menopause. A total of 26 studies met the inclusion and exclusion criteria: 9 studies evaluated the effects of work on andropause symptoms, 8 studies examined its effects on age at menopause onset, and 9 studies addressed its effects on menopausal symptoms. Work-related factors, such as psychological stress, physical effort, and sleep disorders, showed a significant correlation with andropause manifestations, whereas age at menopause and severity of menopausal symptoms were both influenced by factors such as pesticide exposure, high job strain, and repetitive work. Since work accompanies men and women for most of their lives, it is essential to identify and prevent the risk factors that may affect reproductive health.

Opportunities and limits of combining microbiome and genome data for complex trait prediction.

BACKGROUND: Analysis and prediction of complex traits using microbiome data combined with host genomic information is a topic of utmost interest. However, numerous questions remain to be answered: how useful can the microbiome be for complex trait prediction? Are estimates of microbiability reliable? Can the underlying biological links between the host's genome, microbiome, and phenome be recovered? METHODS: Here, we address these issues by (i) developing a novel simulation strategy that uses real microbiome and genotype data as inputs, and (ii) using variance-component approaches (Bayesian Reproducing Kernel Hilbert Space (RKHS) and Bayesian variable selection methods (Bayes C)) to quantify the proportion of phenotypic variance explained by the genome and the microbiome. The proposed simulation approach can mimic genetic links between the microbiome and genotype data by a permutation procedure that retains the distributional properties of the data. RESULTS: Using real genotype and rumen microbiota abundances from dairy cattle, simulation results suggest that microbiome data can significantly improve the accuracy of phenotype predictions, regardless of whether some microbiota abundances are under direct genetic control by the host or not. This improvement depends logically on the microbiome being stable over time. Overall, random-effects linear methods appear robust for variance components estimation, in spite of the typically highly leptokurtic distribution of microbiota abundances. The predictive performance of Bayes C was higher but more sensitive to the number of causative effects than RKHS. Accuracy with Bayes C depended, in part, on the number of microorganisms' taxa that influence the phenotype. CONCLUSIONS: While we conclude that, overall, genome-microbiome-links can be characterized using variance component estimates, we are less optimistic about the possibility of identifying the causative host genetic effects that affect microbiota abundances, which would require much larger sample sizes than are typically available for genome-microbiome-phenome studies. The R code to replicate the analyses is in https://github.com/miguelperezenciso/simubiome .

Automatic Fruit Morphology Phenome and Genetic Analysis: An Application in the Octoploid Strawberry.

Automatizing phenotype measurement will decisively contribute to increase plant breeding efficiency. Among phenotypes, morphological traits are relevant in many fruit breeding programs, as appearance influences consumer preference. Often, these traits are manually or semiautomatically obtained. Yet, fruit morphology evaluation can be enhanced using fully automatized procedures and digital images provide a cost-effective opportunity for this purpose. Here, we present an automatized pipeline for comprehensive phenomic and genetic analysis of morphology traits extracted from internal and external strawberry (Fragaria x ananassa) images. The pipeline segments, classifies, and labels the images and extracts conformation features, including linear (area, perimeter, height, width, circularity, shape descriptor, ratio between height and width) and multivariate (Fourier elliptical components and Generalized Procrustes) statistics. Internal color patterns are obtained using an autoencoder to smooth out the image. In addition, we develop a variational autoencoder to automatically detect the most likely number of underlying shapes. Bayesian modeling is employed to estimate both additive and dominance effects for all traits. As expected, conformational traits are clearly heritable. Interestingly, dominance variance is higher than the additive component for most of the traits. Overall, we show that fruit shape and color can be quickly and automatically evaluated and are moderately heritable. Although we study strawberry images, the algorithm can be applied to other fruits, as shown in the GitHub repository.

A pilot RNA-seq study in 40 pietrain ejaculates to characterize the porcine sperm microbiome.

The microbiome plays a key role in homeostasis and health and it has been also linked to fertility and semen quality in several animal species including swine. Despite the more than likely importance of sperm bacteria on the boar's reproductive ability and the dissemination of pathogens and antimicrobial resistance genes, the high throughput characterization of the swine sperm microbiome remains scarce. We carried RNA-seq on 40 ejaculates each from a different Pietrain boar and found that a proportion of the sequencing reads did not map to the Sus scrofa genome. The current study aimed at using these reads not belonging to pig to carry a pilot study to profile the boar sperm bacterial population and its relation with 7 semen quality traits. We found that the boar sperm contains a broad population of bacteria. The most abundant phyla were Proteobacteria (39.1%), Firmicutes (27.5%), Actinobacteria (14.9%) and Bacteroidetes (5.7%). The predominant species contaminated sperm after ejaculation from soil, faeces and water sources (Bacillus megaterium, Brachybacterium faecium, Bacillus coagulans). Some potential pathogens were also found but at relatively low levels (Escherichia coli, Clostridioides difficile, Clostridium perfringens, Clostridium botulinum and Mycobacterium tuberculosis). We also identified 3 potential antibiotic resistant genes from E. coli against chloramphenicol, Neisseria meningitidis against spectinomycin and Staphylococcus aureus against linezolid. None of these genes were highly abundant. Finally, we classified the ejaculates into categories according to their bacterial features and semen quality parameters and identified two categories that significantly differed for 5 semen quality traits and 13 bacterial features including the genera Acinetobacter, Stenotrophomonas and Rhodobacter. Our results show that boar semen contains a bacterial community, including potential pathogens and putative antibiotic resistance genes, and that these bacteria may affect its reproductive performance.

Proteomic Profiling and Monitoring of Training Distress and Illness in University Swimmers During a 25-Week Competitive Season.

PURPOSE: To evaluate relationships of proteomics data, athlete-reported illness, athlete training distress (TDS), and coaches' ratings of distress and performance over the course of the competitive season. METHODS: Thirty-five NCAA Division II swimmers were recruited to the study (male n = 19, female n = 16; age 19.1 ± 1.6 years). Athletes provided fingerprick dried blood spot (DBS) samples, illness symptoms, and TDS every Monday for 19 of 25 weeks in their season. Coaches monitored performance and rated visual signs of distress. DBS samples were analyzed for a targeted panel of 12 immune-related proteins using liquid chromatography/mass spectrometry (LC/MS). RESULTS: Thirty-two swimmers completed the protocol. The data were grouped in 2-3 weeks segments to facilitate interpretation and analysis of the data. TDS scores varied between athletes, and were highest during the early fall conditioning ramp up period (8.9 ± 1.6 at baseline to a peak of 22.6 ± 2.0). The percent of athletes reporting illness was high throughout the season (50-78%). Analysis of TDS using Principle Component Analysis (PCA) revealed that 40.5% of the variance (PC1) could be attributed to illness prevalence, and TDS scores for the athletes reporting illness and no illness were different across the season (P < 0.001). The coaches' ratings of swim performance and swimmer's distress, sex, and racing distance (sprinters, middle distance, long distance) were not correlated with PC1. Linear Discriminant Analysis (LDA) analysis of the data showed a separation of the baseline weeks from exam weeks with or without competitions, and with competitions alone (p < 0.001). Seven of the 12 proteins monitored over the course of training were upregulated, and the addition of the protein data to LDA analysis enhanced the separation between these groups of weeks. CONCLUSION: TDS and illness were related in this group of 32 collegiate swimmers throughout the competitive season, and expression of immune proteins improved the statistical separation of baseline weeks from the most stressful weeks. TDS data provided by the swimmers did not match their coaches' ratings of distress and swim performance. The importance of the immune system in the reaction to internal and external stress in athletes should be an area of further research.

Estimating Conformational Traits in Dairy Cattle With DeepAPS: A Two-Step Deep Learning Automated Phenotyping and Segmentation Approach.

Assessing conformation features in an accurate and rapid manner remains a challenge in the dairy industry. While recent developments in computer vision has greatly improved automated background removal, these methods have not been fully translated to biological studies. Here, we present a composite method (DeepAPS) that combines two readily available algorithms in order to create a precise mask for an animal image. This method performs accurately when compared with manual classification of proportion of coat color with an adjusted R 2 = 0.926. Using the output mask, we are able to automatically extract useful phenotypic information for 14 additional morphological features. Using pedigree and image information from a web catalog (www.semex.com), we estimated high heritabilities (ranging from h 2 = 0.18-0.82), indicating that meaningful biological information has been extracted automatically from imaging data. This method can be applied to other datasets and requires only a minimal number of image annotations (∼50) to train this partially supervised machine-learning approach. DeepAPS allows for the rapid and accurate quantification of multiple phenotypic measurements while minimizing study cost. The pipeline is available at https://github.com/lauzingaretti/deepaps.

Exploring Deep Learning for Complex Trait Genomic Prediction in Polyploid Outcrossing Species.

Genomic prediction (GP) is the procedure whereby the genetic merits of untested candidates are predicted using genome wide marker information. Although numerous examples of GP exist in plants and animals, applications to polyploid organisms are still scarce, partly due to limited genome resources and the complexity of this system. Deep learning (DL) techniques comprise a heterogeneous collection of machine learning algorithms that have excelled at many prediction tasks. A potential advantage of DL for GP over standard linear model methods is that DL can potentially take into account all genetic interactions, including dominance and epistasis, which are expected to be of special relevance in most polyploids. In this study, we evaluated the predictive accuracy of linear and DL techniques in two important small fruits or berries: strawberry and blueberry. The two datasets contained a total of 1,358 allopolyploid strawberry (2n=8x=112) and 1,802 autopolyploid blueberry (2n=4x=48) individuals, genotyped for 9,908 and 73,045 single nucleotide polymorphism (SNP) markers, respectively, and phenotyped for five agronomic traits each. DL depends on numerous parameters that influence performance and optimizing hyperparameter values can be a critical step. Here we show that interactions between hyperparameter combinations should be expected and that the number of convolutional filters and regularization in the first layers can have an important effect on model performance. In terms of genomic prediction, we did not find an advantage of DL over linear model methods, except when the epistasis component was important. Linear Bayesian models were better than convolutional neural networks for the full additive architecture, whereas the opposite was observed under strong epistasis. However, by using a parameterization capable of taking into account these non-linear effects, Bayesian linear models can match or exceed the predictive accuracy of DL. A semiautomatic implementation of the DL pipeline is available at https://github.com/lauzingaretti/deepGP/.

SeqBreed: a python tool to evaluate genomic prediction in complex scenarios.

BACKGROUND: Genomic prediction (GP) is a method whereby DNA polymorphism information is used to predict breeding values for complex traits. Although GP can significantly enhance predictive accuracy, it can be expensive and difficult to implement. To help design optimum breeding programs and experiments, including genome-wide association studies and genomic selection experiments, we have developed SeqBreed, a generic and flexible forward simulator programmed in python3. RESULTS: SeqBreed accommodates sex and mitochondrion chromosomes as well as autopolyploidy. It can simulate any number of complex phenotypes that are determined by any number of causal loci. SeqBreed implements several GP methods, including genomic best linear unbiased prediction (GBLUP), single-step GBLUP, pedigree-based BLUP, and mass selection. We illustrate its functionality with Drosophila genome reference panel (DGRP) sequence data and with tetraploid potato genotype data. CONCLUSIONS: SeqBreed is a flexible and easy to use tool that can be used to optimize GP or genome-wide association studies. It incorporates some of the most popular GP methods and includes several visualization tools. Code is open and can be freely modified. Software, documentation, and examples are available at https://github.com/miguelperezenciso/SeqBreed.

Identification of rumen microbial biomarkers linked to methane emission in Holstein dairy cows.

Mitigation of greenhouse gas emissions is relevant for reducing the environmental impact of ruminant production. In this study, the rumen microbiome from Holstein cows was characterized through a combination of 16S rRNA gene and shotgun metagenomic sequencing. Methane production (CH4 ) and dry matter intake (DMI) were individually measured over 4-6 weeks to calculate the CH4 yield (CH4 y = CH4 /DMI) per cow. We implemented a combination of clustering, multivariate and mixed model analyses to identify a set of operational taxonomic unit (OTU) jointly associated with CH4 y and the structure of ruminal microbial communities. Three ruminotype clusters (R1, R2 and R3) were identified, and R2 was associated with higher CH4 y. The taxonomic composition on R2 had lower abundance of Succinivibrionaceae and Methanosphaera, and higher abundance of Ruminococcaceae, Christensenellaceae and Lachnospiraceae. Metagenomic data confirmed the lower abundance of Succinivibrionaceae and Methanosphaera in R2 and identified genera (Fibrobacter and unclassified Bacteroidales) not highlighted by metataxonomic analysis. In addition, the functional metagenomic analysis revealed that samples classified in cluster R2 were overrepresented by genes coding for KEGG modules associated with methanogenesis, including a significant relative abundance of the methyl-coenzyme M reductase enzyme. Based on the cluster assignment, we applied a sparse partial least-squares discriminant analysis at the taxonomic and functional levels. In addition, we implemented a sPLS regression model using the phenotypic variation of CH4 y. By combining these two approaches, we identified 86 discriminant bacterial OTUs, notably including families linked to CH4 emission such as Succinivibrionaceae, Ruminococcaceae, Christensenellaceae, Lachnospiraceae and Rikenellaceae. These selected OTUs explained 24% of the CH4 y phenotypic variance, whereas the host genome contribution was ~14%. In summary, we identified rumen microbial biomarkers associated with the methane production of dairy cows; these biomarkers could be used for targeted methane-reduction selection programmes in the dairy cattle industry provided they are heritable.

Gut eukaryotic communities in pigs: diversity, composition and host genetics contribution.

BACKGROUND: The pig gut microbiome harbors thousands of species of archaea, bacteria, viruses and eukaryotes such as protists and fungi. However, since the majority of published studies have been focused on prokaryotes, little is known about the diversity, host-genetic control, and contributions to host performance of the gut eukaryotic counterparts. Here we report the first study that aims at characterizing the diversity and composition of gut commensal eukaryotes in pigs, exploring their putative control by host genetics, and analyzing their association with piglets body weight. RESULTS: Fungi and protists from the faeces of 514 healthy Duroc pigs of two sexes and two different ages were characterized by 18S and ITS ribosomal RNA gene sequencing. The pig gut mycobiota was dominated by yeasts, with a high prevalence and abundance of Kazachstania spp. Regarding protists, representatives of four genera (Blastocystis, Neobalantidium, Tetratrichomonas and Trichomitus) were predominant in more than the 80% of the pigs. Heritabilities for the diversity and abundance of gut eukaryotic communities were estimated with the subset of 60d aged piglets (N = 390). The heritabilities of α-diversity and of the abundance of fungal and protists genera were low, ranging from 0.15 to 0.28. A genome wide association study reported genetic variants related to the fungal α-diversity and to the abundance of Blastocystis spp. Annotated candidate genes were mainly associated with immunity, gut homeostasis and metabolic processes. Additionally, we explored the association of gut commensal eukaryotes with piglet body weight. Our results pointed to a positive contribution of fungi from the Kazachstania genus, while protists displayed both positive (Blastocystis and Entamoeba) and negative (Trichomitus) associations with piglet body weight. CONCLUSIONS: Our results point towards a minor and taxa specific genetic control over the diversity and composition of the pig gut eukaryotic communities. Moreover, we provide evidences of the associations between piglets' body weight after weaning and members from the gut fungal and protist eukaryote community. Overall, this study highlights the relevance of considering, along with that of bacteria, the contribution of the gut eukaryote communities to better understand host-microbiome association and their role on pig performance, welfare and health.

Link-HD: a versatile framework to explore and integrate heterogeneous microbial communities.

MOTIVATION: We present Link-HD, an approach to integrate multiple datasets. Link-HD is a generalization of 'Structuration des Tableaux A Trois Indices de la Statistique-Analyse Conjointe de Tableaux', a family of methods designed to integrate information from heterogeneous data. Here, we extend the classical approach to deal with broader datasets (e.g. compositional data), methods for variable selection and taxon-set enrichment analysis. RESULTS: The methodology is demonstrated by integrating rumen microbial communities from cows for which methane yield (CH4y) was individually measured. Our approach reproduces the significant link between rumen microbiota structure and CH4 emission. When analyzing the TARA's ocean data, Link-HD replicates published results, highlighting the relevance of temperature with members of phyla Proteobacteria on the structure and functionality of this ecosystem. AVAILABILITY AND IMPLEMENTATION: The source code, examples and a complete manual are freely available in GitHub https://github.com/lauzingaretti/LinkHD and in Bioconductor https://bioconductor.org/packages/release/bioc/html/LinkHD.html.

A Guide for Using Deep Learning for Complex Trait Genomic Prediction.

Deep learning (DL) has emerged as a powerful tool to make accurate predictions from complex data such as image, text, or video. However, its ability to predict phenotypic values from molecular data is less well studied. Here, we describe the theoretical foundations of DL and provide a generic code that can be easily modified to suit specific needs. DL comprises a wide variety of algorithms which depend on numerous hyperparameters. Careful optimization of hyperparameter values is critical to avoid overfitting. Among the DL architectures currently tested in genomic prediction, convolutional neural networks (CNNs) seem more promising than multilayer perceptrons (MLPs). A limitation of DL is in interpreting the results. This may not be relevant for genomic prediction in plant or animal breeding but can be critical when deciding the genetic risk to a disease. Although DL technologies are not "plug-and-play", they are easily implemented using Keras and TensorFlow public software. To illustrate the principles described here, we implemented a Keras-based code in GitHub.