Bioinformatics
Department of Informatics and Chemistry,
Faculty of Chemical Technology
Deep learning for computational drug development
Svozil Daniel doc. Mgr. Ph.D.
Annotation
Artificial neural networks had their first heyday in drug discovery approximately two decades ago. Currently, we are witnessing renewed interest in adapting advanced neural networks for pharmaceutical research by borrowing from the field of "deep learning". The aim of deep neural networks is to extract the underlying patterns and important features hidden beneath the immense complexity of data using sophisticated multilayer architectures. However, compared with other life sciences, the application of deep neural networks in drug discovery is still limited. The aim of the dissertation is to investigate the applicability of deep neural networks in various computational drug development tasks, such as target identification or biomarker discovery.
Selection Pressure in Human Endogenous Retroviral Elements (HERVs)
Pačes Jan Mgr. Ph.D.
Annotation
Retroviral elements comprise a significant portion of the human genome. They can be transribed in certain tissues, however their effect on the cell fenotype is yet unknown. Analysis of selection pressure on individual HERVs loci may shed light on their biological effects. The dissertation is aimed at a computational analysis of this pressure based on publicly available large scale data (1000 genomes project). The results will be integrated with the already existing HERVd database (http://herv.img.cas.cz). Part of the project is also design of visualization of selection pressure and integration into current web interface.
Deep learning in chemoinformatics and bioinformatics
Svozil Daniel doc. Mgr. Ph.D.
Annotation
In the big data era, voluminous datasets are routinely acquired, stored and analyzed with the aim to inform biomedical discoveries and validate hypotheses. No doubt, data volume and diversity have dramatically increased by the advent of new high throughput experimental technologies such as DNA/RNA sequencing or screening of biological activity. The identification and interpretation of relationships in such complex data then requires the use of state-of-the-art data mining approaches, in which deep learning methods play a prominent role. The dissertation will focus on the application of deep learning methods to various chemoinformatics and bioinformatics problems.
Methods of Single Cell RNA-seq Data Processing
Svozil Daniel doc. Mgr. Ph.D.
Annotation
Single-cell RNA-sequencing (scRNA-seq) has emerged in the past decade as a new tool for simultaneously accessing the transcriptome of thousands of individual cells and is becoming a standard element in the toolbox of molecular biologists. The new-born field of single-cell transcriptomics (or single-cell genomics, in general) is evolving rapidly. This massive technological progress has been accompanied by the development of numerous algorithms for analyzing scRNA-seq data. These analytic tools are designed to overcome key challenges that originate from the substantial technical variability and sparseness inherent in scRNA-seq data. The aim of this dissertation is to investigate the applicability of different scRNA-seq analytic tools in various pathophysiological projects focusing on discovering function of cells in CNS. Moreover, the dissertation will aim to define standardized pipelines for scRNA-seq data analysis.
Biotechnologický ústav AV ČR, v.v.i.
Signaling pathways and functional changes in malignancies
Kolář Michal Mgr. Ph.D.
Annotation
Activity of signaling pathways creates an intercellular image of the extracellular environment. This model is used to create appropriate response, which involves, in majority of the cases, changes in transcriptional activity of target genes and changes in their expression. Defects in the signaling pathways are common in malignant and other diseases and lead to aberrant responses to extracellular signals and in turn to pathological states. A prominent example is dismal activation of the ERK-signaling pathway caused by specific mutations in malignant melanoma, which leads to uncontrolled tumor growth. The activity of signaling pathways can be well estimated from the whole-genome transcriptional activity. The proposed work will focus on mutual interaction of cancer cells and tumour microenvironment, signaling pathways, on description of the trancriptional activity at whole genome scale, statistical analysis of the data, and biological interpretation of the observed changes with the aim to identify new clinical markers and/or therapeutical targets.