The identification of social interactions is of fundamental importance for animal behavioral studies, addressing numerous problems like investigating the influence of social hierarchical structures or the drivers of agonistic behavioral disorders. However, the majority of previous studies often rely on manual determination of the number and types of social encounters by direct observation which requires a large amount of personnel and economical efforts. To overcome this limitation and increase research efficiency and, thus, contribute to animal welfare in the long term, we propose in this study a framework for the automated identification of social contacts. In this framework, we apply a convolutional neural network (CNN) to detect the location and orientation of pigs within a video and track their movement trajectories over a period of time using a Kalman filter (KF) algorithm. Based on the tracking information, we automatically identify social contacts in the form of head–head and head–tail contacts. Moreover, by using the individual animal IDs, we construct a network of social contacts as the final output. We evaluated the performance of our framework based on two distinct test sets for pig detection and tracking. Consequently, we achieved a Sensitivity, Precision, and F1-score of 94.2%, 95.4%, and 95.1%, respectively, and a $MOTA$ score of 94.4%. The findings of this study demonstrate the effectiveness of our keypoint-based tracking-by-detection strategy and can be applied to enhance animal monitoring systems.

Machine learning is frequently used to make decisions based on big data. Among these techniques, random forest is particularly prominent. Although random forest is known to have many advantages, one aspect that is often overseen is that it is a non-deterministic method that can produce different models using the same input data. This can have severe consequences on decision-making processes. In this study, we introduce a method to quantify the impact of non-determinism on predictions, variable importance estimates, and decisions based on the predictions or variable importance estimates. Our findings demonstrate that increasing the number of trees in random forests enhances the stability in a non-linear way while computation time increases linearly. Consequently, we conclude that there exists an optimal number of trees for any given data set that maximises the stability without unnecessarily increasing the computation time. Based on these findings, we have developed the R package optRF which models the relationship between the number of trees and the stability of random forest, providing recommendations for the optimal number of trees for any given data set.

Background: The ever-increasing availability of high-density genomic markers in the form of single nucleotide polymorphisms (SNPs) enables genomic prediction, i.e. the inference of phenotypes based solely on genomic data, in the field of animal and plant breeding, where it has become an important tool. However, given the limited number of individuals, the abundance of variables (SNPs) can reduce the accuracy of prediction models due to overfitting or irrelevant SNPs. Feature selection can help to reduce the number of irrelevant SNPs and increase the model performance. In this study, we investigated an incremental feature selection approach based on ranking the SNPs according to the results of a genome-wide association study that we combined with random forest as a prediction model, and we applied it on several animal and plant datasets. Results: Applying our approach to different datasets yielded a wide range of outcomes, i.e. from a substantial increase in prediction accuracy in a few cases to minor improvements when only a fraction of the available SNPs were used. Compared with models using all available SNPs, our approach was able to achieve comparable performances with a considerably reduced number of SNPs in several cases. Our approach showcased state-of-the-art efficiency and performance while having a faster computation time. Conclusions: The results of our study suggest that our incremental feature selection approach has the potential to improve prediction accuracy substantially. However, this gain seems to depend on the genomic data used. Even for datasets where the number of markers is smaller than the number of individuals, feature selection may still increase the performance of the genomic prediction. Our approach is implemented in R and is available at https://github.com/FelixHeinrich/GP_with_IFS/ .

Faba bean (Vicia faba) is a grain legume, which is globally grown for both human consumption as well as feed for livestock. Despite its agro-ecological importance the usage of Vicia faba is severely hampered by its anti-nutritive seed-compounds vicine and convicine (V+C). The genes responsible for a low V+C content have not yet been identified. In this study, we aim to computationally identify regulatory SNPs (rSNPs), i.e., SNPs in promoter regions of genes that are deemed to govern the V+C content of Vicia faba. For this purpose we first trained a deep learning model with the gene annotations of seven related species of the Leguminosae family. Applying our model, we predicted putative promoters in a partial genome of Vicia faba that we assembled from genotyping-by-sequencing (GBS) data. Exploiting the synteny between Medicago truncatula and Vicia faba, we identified two rSNPs which are statistically significantly associated with V+C content. In particular, the allele substitutions regarding these rSNPs result in dramatic changes of the binding sites of the transcription factors (TFs) MYB4, MYB61, and SQUA. The knowledge about TFs and their rSNPs may enhance our understanding of the regulatory programs controlling V+C content of Vicia faba and could provide new hypotheses for future breeding programs.

Transcription factors (TFs) are a special class of DNA-binding proteins that orchestrate gene transcription by recruiting other TFs, co-activators or co-repressors. Their combinatorial interplay in higher organisms maintains homeostasis and governs cell identity by finely controlling and regulating tissue-specific gene expression. Despite the rich literature on the importance of cooperative TFs for deciphering the mechanisms of individual regulatory programs that control tissue specificity in several organisms such as human, mouse, or Drosophila melanogaster, to date, there is still need for a comprehensive study to detect specific TF cooperations in regulatory processes of cattle tissues. To address the needs of knowledge about specific combinatorial gene regulation in cattle tissues, we made use of three publicly available RNA-seq datasets and obtained tissue-specific gene (TSG) sets for ten tissues (heart, lung, liver, kidney, duodenum, muscle tissue, adipose tissue, colon, spleen and testis). By analyzing these TSG-sets, tissue-specific TF cooperations of each tissue have been identified. The results reveal that similar to the combinatorial regulatory events of model organisms, TFs change their partners depending on their biological functions in different tissues. Particularly with regard to preferential partner choice of the transcription factors STAT3 and NR2C2, this phenomenon has been highlighted with their five different specific cooperation partners in multiple tissues. The information about cooperative TFs could be promising: i) to understand the molecular mechanisms of regulating processes; and ii) to extend the existing knowledge on the importance of single TFs in cattle tissues.

The interactions between SNPs result in a complex interplay with the phenotype, known as epistasis. The knowledge of epistasis is a crucial part of understanding genetic causes of complex traits. However, due to the enormous number of SNP pairs and their complex relationship to the phenotype, identification still remains a challenging problem. Many approaches for the detection of epistasis have been developed using mutual information (MI) as an association measure. However, these methods have mainly been restricted to case–control phenotypes and are therefore of limited applicability for quantitative traits. To overcome this limitation of MI-based methods, here, we present an MI-based novel algorithm, MIDESP, to detect epistasis between SNPs for qualitative as well as quantitative phenotypes. Moreover, by incorporating a dataset-dependent correction technique, we deal with the effect of background associations in a genotypic dataset to separate correct epistatic interaction signals from those of false positive interactions resulting from the effect of single SNP×phenotype associations. To demonstrate the effectiveness of MIDESP, we apply it on two real datasets with qualitative and quantitative phenotypes, respectively. Our results suggest that by eliminating the background associations, MIDESP can identify important genes, which play essential roles for bovine tuberculosis or the egg weight of chickens.