Cancer Systems Biology

Cancer Systems Biology

Our group (Cancer System Biology, CSB) applies a system-biology view to the analyses of different sources of cancer data (e.g., gene expression, genomics, proteomics) to identify biomarkers. In addition, we complement our studies with structure-based approaches to classify cancer variants or to study interactions between cancer proteins and small molecules. Our research is applied both Pan-cancer than to specific cancer types, with the main focus on childhood and breast cancers. The group activities have been developed naturally in a translational direction, with the basic research on cancer mechanisms being the base for the more applicative and translational one.


Parts of the activities are carried out in close collaboration with clinicians in Denmark and Sweden thanks to our participation in the i-COPE Consortium ( 

We are also affiliated with the Danish Cancer Society Research Center with the Computational Biology Laboratory (CBL,, where our colleagues focus on structural biology and experiments with biochemical and cellular assays. The CSB and CBL groups work closely together with the results generated by CSB on interesting cancer targets analyzed under the lens of structural biology at CBL. CBL can also support experimental validations with biochemical assays in vitro or cellular models.

We also work closely with the University of Miami (Prof. Steven Chen and Dr. Antonio Colaprico) to develop and maintain software for the analyses of large cancer datasets.


Prediction of cancer drivers and biomarkers

We have been developing Moonlight, a framework based on the integration of different biological data to identify driver genes in cancer and classify them in tumor suppressors and oncogenes. The framework is also used to identify dual role genes, i.e., those genes that can act both as tumor suppressors or oncogenes depending on the cellular context. 




The research that we carry out with Moonlight is accompanied by other approaches for clustering and stratification of the samples depending on their gene expression, genomics, or proteomics signatures. As an example, we have been developing CAMPP (CAncer bioMarker Prediction Pipeline,  for the analyses of different sources of cancer data. CAMPP includes data-management tools for missing value imputation, quality checks, normalization protocols, data clustering, differential expression, survival, and network-based analyses.

Classification and prioritization of cancer variants

We aim to provide an atlas of cancer alterations affecting the protein products that can also be consulted to prioritize variants for experimental or functional studies.

We aim to go beyond the available tools to predict pathogenic mutations based on sequence or simple structural features. These methods do not provide enough details about the mechanisms related to the alterations and how we can counteract them.

In our workflow, we aim to comprehensively account for different aspects that mutational events alter in terms of protein structure, dynamics, and function. For examples, the variants are ranked and classified according to the alteration at the level of stability of the protein product, changes in post-translational regulations, interactome, also including allosteric changes that are the results of the alteration at a distal site with respect to the functional one. As an example, we applied the workflow to the study of a key autopaghy protein and its cancer-related mutations.


Germline-Somatic Continuum in pediatric cancer 

In this work, we explore the emerging concept of the Germline-Somatic continuum using Next Generation Sequencing (NGS) data obtained from cohorts of Scandinavian children diagnosed with cancer. According to the Germline-Somatic continuum view, low-penetrance germline mutations cooperate with a number of somatic mutations and they can have additive effects, ultimately inducing cancer onset and development.  To this goal, we have been analyzing germline, tumor, and somatic variants by Whole Genome Sequencing (WGS), RNAseq, and WGS+RNAseq, respectively. We aim to provide a framework to make sense of cancer genomic data and identifying the specific damaged pathways and molecular dependencies towards precision medicine. This research area is carried out  in close collaboration with the Disease Data Intelligence (DDI) group in the Section.

Machine learning for pharmacogenetics 

In collaboration with different Danish hospitals, we also aim to apply advanced computing and artificial intelligence to understand and describe the pharmacogenetic alterations in childhood cancer patients. With over forty chemotherapeutic drugs approved in the current acute lymphoblastic leukaemia treatment protocol, it is estimated that 15% of all patients have some form of pharmacogenetic interaction. We aim to develop reliable risk assessment models for leukaemia treatment that predict drug response and toxicities by integrating clinical, genetic and biological data. 



Elena Papaleo
DTU Health Tech