Chemoinformatics and Imaging

Background

Our work group aims at providing IT based solutions for biochemical aspects of the systems biology and systems medicine projects.

We develop and adapt software solutions to provide the required tools for medical research.
We accomplish this by applying and extending bioinformatics, cheminformatics and image processing methods to complement systems biology research questions integrating the means for statistical analysis. While extending our methodology toolbox we explicitly try to strengthen the planed analysis by integrating early on the requirements for the statistical analysis of the output of our pipelines.

Head of the group: Dr. Nietert

As more and more image based research is conducted and the number of created images per project increases, we develop the solutions for an automated analysis, e.g. AutoBuSTeD project, but if necessary can even venture out to help improve the acquisition setup.

Research foci

For the past years we focus our research in the field of cystic fibrosis, where we still see an unmet demand for IT solutions to help this field.

Our CandActCFTR project aims at providing means to distil and curate compound libraries from literature and other online sources and combine this structured information for analysis.

CF is a good target for a system medicine approach

At a first glance, CF is a monocausal disease in which over 2000 putative mutations leading to various forms of phenotypes have been identified. Among these, about 300 variants define the more common types. The CFTR protein is only effective as an integral membrane protein, and as such, it is affected by transcription, translation, folding and degradation, as well as protein traficking processes. Thus, this monogenic disease has multiple sites for potential drug intervention during its life cycle and covers also protein structure variants. This makes it an interesting target for a system medicine approach. The life cycle of the protein offers also multiple modes to obtain information to annotate the system and the existing literature offers various annotations for specific combinations of mutations and read-outs (e.g. protein expression, functional patch clamp measurements, up to structure models and molecular dynamic simulations).

Current thesis offers

In the context of the project CandActCFTR work is possible at all levels (internship, bachelor and master). Possible works could be:

1.       CandActBase MIRIAM Module - Minimal Information Required in the Annontation of  Models

  • Implementation of the automated storage and retrieval of Systems Biology Markup Language SBML models - storage, retrieval, updating/extension of annotations and merging of models
  • working on Interactive visualization for Systems Model depictions
  • conducting comparative statistics analysis

2.          CandACtBase Protein DataBank/ Molecular Dynamics Trajectory Module

  • Automated Target structure model annotation - includes work on storage, retrieval updating/ extension of annotations in the software context of CandAct
  • Visualization - e.g. for preparation for virtual docking and linked storage of results, binding site visualization, comparative statistics using similarity descriptions. Developing search modules interfaces and report views.

3.          CandActBase SystemsBiology Curation and Annotation Helper Module(s)

           

3.1 Topics modelling

  • Automated Paper Information retrival and model annotation - Testing if using topics modelling for text sources can help curate a data base like CandActCFTR
  • Testing how similar the flags can be derived from the literature compared to manual gold standard
  • Automatically extracting topics from the paper stacks and using this to annotate the data
  • Providing links to the source positions - Text mining - to extend the annotion network

 

3.2 Chemical Information retrieval from PDFs

  • Extracting all individual chemical names from the papers and if possible also if available the chemical structures (molecular graphs) from images - integration of existing libraries  like OSRA: Optical Structure Recognition
  • Automating lookup of structures from chemical names

 

In the context of the image analysis projects from Manuel Nietert's group, work is possible at all levels (internship, bachelor and master). Possible works could be:

 

4. Imaging related (possible data sources: e.g. CT, MRI, microscopy, ultrasonic)

  • Image Analysis, e.g. detection of objects, quantification, tracing, fusion
  • 3D reconstruction and visualization, e.g. from laser scanning microscopy
  • developing pipelines to be used later by non-experts: includes GUI development

 

e.g.

4.1 Automated Lung volume tracing in real time MRI time series diagnosing Pompe

  • visualization of results via data fusion and augmentation: geometry, vector  and matrix math
  • 3D reconstruction
  • optional deep learning applied for optimized organ detection / segmentation
  • quantification and statistics
  • pipeline and GUI

 

4.2 QuantOS – Quantification of Organoid Swelling in Cystic Fibrosis – with University Verona, Italy

  • detection of organoids in microscopy image time series
  • quantification and statistics
  • pipeline and GUI

4.3 QuantOG – Quantification of Organoid Growth in Cancer Models - with UMG

  • detection of organoids in microscopy image time series
  • quantification and statistics
  • pipeline and GUI

4.4 3D reconstruction from a StereoCameraSystem and overlay with CT data

  • visualization of results via data fusion and augmentation: geometry, vector  and matrix math
  • 3D reconstruction
  • detection / segmentation
  • quantification and statistics

pipeline and GUI

 

We have a constant input of new projects from our collaboration partners, sometimes with hardware optimisation tasks as well:

 

5. Automated optimization of the lighting for a diagnostic test

The bubble sweat test to diagnose cystic fibrosis is based on capturing images of the skin and quantifiying the sweat bubble growth behaviour. As the bubbles are detected by their reflection edges in the images optimal lighting is crucial for the robust application. In order to improve the acquisition quality for future measurements we strife to improve this now further.

  • establish the control of the ring light by a microcontroller
  • integration of image analysis to control the light intensity to optimize the parameter
  • geometrical optimization of the setup
  • optional: color optimization using different light sources / wave length

Members of the group

Dr. Manuel M. Nietert (head)
Liza Vinhoven (PhD student)
David Tschritter (Bachelor student)
Paul Wuerzberg (Bachelor student)

Former Members of the group

More soon…

Open thesis positions

We are currently seeking candidates to join our group at any of the following levels (Bachelor thesis, Master thesis).

Please contact:

Group Leader

Dr. Manuel Nietert

contact information

Follow us