Application of single cell sequencing and machine learning in mammary gland biology

The mammary gland is a highly specialised organ in mammals, which has an extremely important role in reproduction and is essential for economical milk production in agriculture. The capacity for milk production in dairy cows exceeds several times the nutritional needs of the calf and represents a unique production trait, which has been efficiently improved using the classical selection approach. Therefore, lactation in cattle is an ideal model for studying the biology of lactation with the aim to discover the mechanistic base of this complex trait at cellular level with the potential to contribute to the basic knowledge about lactation biology. Recently, it has become possible to investigate the single cell transcriptomes instead of bulk RNA pools from different cell types. Since then, mammary epithelial cells at the single-cell level in humans and mice were examined and revealed much higher heterogeneity of the mammary epithelial cell population than previously reported. To date, no experiment has been carried out to produce the profile of bovine mammary gland gene expression using the single cell RNA sequencing (scRNA-Seq) approach, nor has milk transcriptome been profiled at the single cell level. Within the frame of this project, scRNA-Seq will be applied to document cell type specific expression profiles in the mammary gland and to determine different cell types based on cell type specific transcription profile. This approach will allow us to identify cellular sources for several milk components, which did not have a defined origin before. Further analysis of transcriptomic data will allow for the identification of regulatory elements (transcription factors, prediction of binding sites for mammary gland expressed transcription factors, etc.). With the application of inter species comparison of transcriptomic profiles we will try to identify generally and species -specific expressed genes. Single cell transcriptomics data characterizes living systems at an unprecedented level of resolution, however, it is extremely sparse and noisy. As scRNA-seq data has potential to reveal novel insights into complex biological systems, it also poses several new algorithmic challenges. We will apply machine learning methods to address the problems related to scRNA-Seq data analysis and integration of transcriptomic data with the chromatin structure data provided by scATAC-Seq.