Topic and research project: Specialist for propensity score matching, modelling and solving Mixed-Integer-Programs which arise in bioinformatic problems of the group, especially on networks and machine learning, statistics. System administration of the group. Methods for propensity matching and machine learning are implemented in a project, in which we are developing a rule for macrolide /beat-lactam combination therapy in community-acquired pneumonia (CAP) of moderate severity.
Dr. Thomas Beder
Topic: Employing machine learning for remodeling the Anopheles gambiae immune response to Plasmodium falciparum infection
Research project: Malaria, caused by the Plasmodium parasite, affects approximately 3 billion people worldwide each year. In sub-Saharan Africa, the major vector for Plasmodium falciparum is the female Anopheles gambiae mosquito. Not only humans but also the vector is damaged by malaria and an effective and elaborated immune response has evolved. We use systems biology approaches to find new targets that confer resistance to A. gambiae preventing its vector capability. Besides this, machine learning methods are applied to identify patient variables highly correlating with death after hospitalization of community-acquired pneumonia (CAP) of moderate severity.
Dr. Daniela Röll
Topic and research project: Host response to infectious diseases, single cell sequencing and identification of specific gene regulators.
Dr. Olga Talkenberg
Topic: Germ detection by UV-light excited auto-fluorescence
Research project: Benefiting from the new UV-LED generation, UV-light excited fluorescence becomes a very powerful tool in life sciences. We are developing a new handheld device for direct germ detection in a very short timeframe based on auto-fluorescence of the germs. The project “GermDetect” belongs to the BMBF supported ‘Advanced UV for life’ consortium.
Dr. Vasily S. Romanov
Topic and research project: Data mining is a powerful contemporary approach in solving various questions. Analysis of large datasets by machine learning, in particular, helps to find hidden patterns of data and to build statistical models, which are used for describing new data and predicting outcomes of corresponding processes. Such methods are utilized in several of our projects, which deal with different types of data. Applying machine learning algorithms in analysis of Excitation Emission Matrices (EEMs) of fluorescence spectra allows us to distinguish even small variations in light intensity. This approach is successfully used on our experimental dataset of the “GermDetect” project, which aims at detection by UV autofluorescence of very small amounts of biological contaminations. Besides, data mining methods are applied to datasets of patients with community-acquired pneumonia (CAP) in order to find the best antibiotic combination for the treatment, which depends on various patient factors and medical parameters.
Dr. Oyelade Olanrewaju Jelili
Topic: Deciphering the protein-protein interaction network of Anopheles gambiae: A computational approach based on machine learning
Research Project: Malaria tropica is the most severe form of malaria, and particularly in sub-Saharan African countries a life threatening central health care problem. The disease is mainly transmitted by the mosquito Anopheles gambiae. The project aims to find targets for insecticide treatment tipping particularly the infected mosquitoes. Protein-protein interactions play an important role in the prediction of protein function and the prediction of a target protein for insecticides. However, there exists only very limited information about protein-protein interactions for A. gambiae. This project aims to develop computational models to infer protein-protein interactions of Anopheles gambiae using protein-protein interaction information of Drosophila melanogaster and other model organisms.
Topic: Predicting nutritional uptakes of Bacillus subtilis and Plasmodium falciparum by integrating gene expression profiles into constrained based metabolic models
Research project: Uptake of nutrition is essential for every organism. Hence, these pathways may suit as therapeutic targets tipping invading pathogens into the host. To implement the method, we are developing flux balance analysis (FBA) based models using the stoichiometric knowledge of the metabolic reactions of a cell, 13C metabolic flux data and transcriptomic data from the model organism B. subtilis. We aim to predict nutritional uptakes only basing on the transcription profiles. This approach can then also be used in more complex settings, such as for investigating the nutritional uptake in opportunistic pathogenic microorganisms in host cells, such as P. falciparum in the red blood cell, in which 13C analysis is difficult.
Topic: Germ detection by UV auto-fluorescence
Research project: We are establishing standard operating procedures to implement and validate a UV-LED based germ detection device.
Topic: Identifying host factors as drug targets for SARS-CoV-2 from RNAi knockdown screens
Research Project: A novel coronavirus was recently discovered and termed SARS-CoV-2. Patients with COVID19 can worsen in a short time-lapse and die of multiple organ failure. Until now, there is no established therapy for COVID–19. A promising alternative can be to repurpose clinically approved drugs. The goal of the project is to employ data from a genome-wide (druggable) RNAi knockdown screen + proteomics profiles of SARS-CoV-2 infected cells + qualitative phenotyping microscopy images to elucidate host-dependency factors that are needed for virus entry into the cell, replication, and spread. A cellular network with host dependency factor modules will allow us to target them with one drug or combination of drugs.
Topic and research project: Examine the correlation between the mutation of FMS-like tyrosine kinase (FLT3) and aberrant bone development in acute myeloid leukemia (AML) patients by a gene expression analysis of bone marker genes in FLT3 ITD vs FLT3 WT AML samples.