Contact informationPlease contact Eivind Hovig for more information
Creating Bioinformatic Tools to automate pathological analysis
Eivind Hovig´s research group works towards putting bioinformatics to work in various aspects of cancer research, with the aim of impacting clinical cancer practices of precision medicine. This is achieved in various ways.
Digital analysis of cancer brings challenges in handling and processing large data sets. The human body contains 150-200 different cell types, and we are outlining which genes are expressed in each of these in order to detect which genes are aberrant and possibly linked to cancer development. Artificial intelligence and decomposing complex signals are key elements to our approach. We use publicly available research databases to test our methods and verify them with pathologists.
High-throughput DNA sequencing is mediating a revolution within both diagnostics, prognostics and treatment options for most cancer types. All cancers seem to be essentially caused by alterations in the DNA in one or more body cells that are then propagated to develop a tumor that then can spread to form metastases in other parts of the body. Different alterations lead to different effects in different cell types, and a main thrust in international cancer research is to identify and quantify these alterations, and eventually to develop drugs to fight the cancer consequences of these alterations.
Understanding mutations in cancer
One main line of research is the identification of the single-base DNA variants that are altered in cancer cells. A main class of the single-base DNA variants occur in the DNA that encodes proteins, i.e. the coding variants. It is this class that is currently a main focus of drug-development, as they may have directly observable consequences on cellular functions. We are providing the full pipeline of DNA exome sequence analysis, along with information on the consequences of each of the millions of observed mutations in human cancer. We are also working to develop this further towards providing all results in understandable formats for both the clinician and the patient. A more difficult class of variants are the small insertions and deletions of DNA material, as well as copy-number variation. These classes of mutations are more difficult to interpret, as they are less well-known and characterized. We are therefore establishing optimized analyses for these classes.
Targeted drug predictions
As progressively more and more mutations with cancer relevance become known, with parallel efforts of targeted drug development, the more drugs are developed. As many drugs target genes and proteins with different cellular functions, it is increasingly possible to treat a cancer with a combination of drugs. However, given a modest number of mutations in a patient, and given a set of drugs for those mutations, it very rapidly becomes very difficult to select the most efficient combination of drugs, as each tumor very often has a unique combination of mutations. We are therefore actively involved in pursuing options to predict combinatorial drug response based on existing clinical and preclinical data. Further, there is currently an unmet need to feed the outcome of the precision medicine drug selections used in the clinic back into the clinical systems for learning the patterns of response.
A completely different level of information with respect to cancer and mutations is that of the organization of DNA in three dimensions in the cell nucleus. We have in recent year published methods to analyze lab-generated data that shed light on the 3D organization, and we have recently in collaboration with scientists as SISSA in Trieste, Italy developed a full three-dimensional model of how the nuclear DNA is organized. When we analyze the results of this modeling, we find that a number of core features of human DNA is actually reproduced in the model. A video model of the organization is shown here: https://youtu.be/bS2cX1of35c
We are now pursuing solutions to model existing cancer 3D data in combination with information on where mutations occur, to further the understanding of how elements that can occur together in 3D space can cause or modulate cancer. For instance, a number of nuclear functions that impact mutational processes have been described.