SBMC 2016 Munich – 6th Conference on Systems Biology of Mammalian Cells

From April 6th to April 8th 2016, the 6th Conference on Systems Biology of Mammalian Cells (#SBMC2016) took place at Klinikum rechts der Isar in Munich, Germany. While you can find the official program on the conference’s webpage, I here picked some of my favorite topics, summarize them and refer to the relevant resources. Bottom line is: Biology is not enough…

Already during the welcome address, host Fabian Theis made clear that systems biology is complex. There is multiple layers from single atoms up to populations of human individuals in a patient cohort. When it comes to mammalian cells, multiple omics levels have to be integrated preferably at the single-cell level to address heterogeneity, which is challenging by itself. Additionally, Fabian strengthened that systems biology needs to be tailored to address emanating questions concerning human health to path the way towards personalized (systems) medicine. Representatives of the German Federal Ministry of Education and Research underlined this strategy by pointing out the e:Med research and funding concept.

Session: Image-based Systems Biology

The first invited talk of was given by Robert Murphy. He aims at generating probabilistic models of cell, nucleus and organelle shape. Software is available under and The latter project tries to create dimensional representations of objects, given their individual distances to each other. Underlying R functions can be applied not only to American cities but also to cells in a tissue slide.

An outstanding presentation selected from the submitted abstracts was held by the physicist Fabian Rost on tail regeneration of axolotl. Fabian elegantly showed by Bayesian inference that proliferation within a zone close to the amputation area is responsible for experimentally observed outgrowth during regeneration. As they have seen a change of cell cycle durations and a kink in outgrowth length 3 to 4 days upon amputation, it seems likely to me that some bang-bang optimality control applies here.

MTZ Award

The MTZ foundation awards prizes to young investigators for their dissertations with exceptional relevance in medical systems biology. This year’s three awardees were:

Plenary Talk

Chris Sanders advised the audience to develop decision-support systems that, based on mathematical models, help clinicians to deliver the right treatment to the right patients. Mentioned aspects about integration of diverse data sets reminded me of a recent Nature comment. Parts of developed software on evolution of protein structures are available through the Debora Marks Lab.

Maria Polychronidion, one of the editors from Molecular Systems Biology, afterwards outlined the editorial process of the journal at EMBO press. She informed us that it is now possible to submit work directly from the pre-print server bioRxiv and encouraged authors to provide source data under: to meet the endeavour to make science more transparent and more accessible.

Session: Single-Cell Systems Biology

Bernd Bodenmiller showed to us in the first invited talk of this session how stunning data obtained by image mass cytometry looks like. As demonstrated in a 2014 Nature Methods paper, tissues are atomized into particle clouds while keeping individual cells in shape.

Principle of image mass cytometry from

In two of the subsequent oral presentations, important questions were raised. Edith Ross from the Markowetz lab asked if tumor evolution can be inferred without a-priori knowledge about clonal lineage trees from single-cell sequencing data. The software oncoNEM (R source code freely available on Bitbucket) is a probabilistic method based on the nested structure of mutations between cells. But how reliable are present single-cell sequencing methodologies? The group of Wolfgang Enard systematically compared different RNA-sequencing methods. The cheapest preparation allows Drop-seq with approximately 0.1€ per library while SCRB-seq is most cost-efficient per experiment. All results were made available as pre-print on bioRxiv.

Session: Signaling Modeling

In the first invited talk of the Signaling Modeling session, Alexander Hoffmann introduced the concepts behind their Science paper from 2014. I read this report right after it was published. In the following, I present the key points taken from my notes back then:

  • Potential information loss in signaling networks through intrinsic noise (i.e. stochasticity inherent to reaction rate constants) and extrinsic noise (i.e. variability in cellular protein abundances).
    • Intrinsic noise: mean of squared error between two successive trajectory points.
    • Extrinsic noise: mean of squared error between single trajectory and average.
  • Information transmission capacity in time-resolved dose-response signaling data encoded in:
    • maximum response amplitude,
    • maximum response time,
    • maximum slope of dynamic response, and
    • ratio between initial and maximum response amplitude.
  • Determination of uncertainty derived from purely extrinsic noise by multiple time-resolved measurements.
  • Analyzed mutual information between ligand and signaling response with varying signal-to-noise ratios revealed that dynamic responses mitigate both intrinsic and extrinsic noise.

Session: Metabolism

From the University of Jena, Germany, Ina Bergheim was invited to talk about ethanol metabolism in the human liver. She noted that even in people, like me, who never drink alcohol, intestinal bacteria produce it. I rather thought ethanol is produced by the detoxifying enzymes of our own body in absence of EtOH, but I might be mistaken. Ina linked ethanol uptake to metabolic changes of the NAD+/NADH ratio and enzymes of the cytochrome P450 family. When I saw the data, a very apparent question came to my mind: Are they no physiology-based pharmacokinetic/pharmacodynamic models for alcohol consumption? A quick PubMed search only listed studies on drug actions altered by alcohol intake but not the effect of ethanol alone.

The bang-bang optimality control came into my mind again as Sandro Hutter spoke about dynamic glycosylation flux analysis. While conventional flux balance analyses rely on steady-state assumptions and are therefore static, pioneering work of Maciek Antoniewicz et alii introduced the notion of dynamics in metabolic networks (c.f. review in Current Opinion in Biotechnology 2013). Interestingly, Sandro observed switching times in metabolic fluxes and explained them by a transiently changing behavior of the cells. In bang-bang optimality control, the line of argument is exactly reversed saying that cells change their behavior because of altered signaling and metabolic processes. Cause and effect have to be disentangled not to end up in tautology…

Session: Multi-Scale Approaches

The invited talk by Markus Löffler was very impressive. The topic was chemotherapy-induced haematotoxicity. First, he outlined a regulatory window. In this range of a sigmoidal dose-response curve for a certain drug, clinical outcome (e.g. granulocyte count) can be tuned without harmful under- or overdosing. Markus strengthened that developed mathematical models need to be relevant for clinical use. It is insufficient to find a mathematical description of a biomedical case if patients do not benefit from this model. I still remember when I first read the PLoS ONE paper from 2013 where Markus and his colleagues describe “A Biomathematical Model of Human Erythropoiesis under Erythropoietin and Chemotherapy Administration”. We used this work as an inspiration for our project on steering erythropoiesis ex vivo.

The second invited talk of this session was given by Alexander Anderson. Sandy made a encouraging plea for the collaboration between biologists, clinicians and theoreticians. He is hosting an annual workshop on Integrative Mathematical Oncology (IMO), which creates a fruitful working environment to bring all the experts together synergistically. The success of such events is underlined by a recent publication emanating from one of the workshops that appeared in journal Cancer Research.

Session: Systems Medicine

The recurring idea(l)s of the conference cumulated in the selected presentation by Fabian Fröhlich. In his project, they wanted to establish predictive models for personalized medicine. The aim to identify, as Fabian put it, “the right drug for the right patient” sounds simple but is challenged by the complexity of genomic data. They used data from the Broad-Novartis cancer cell line encyclopedia and build genome-scale mechanistic dynamic models. To facilitate simultaneous estimation of thousands of model parameters, new methods were developed for the reduction of computational costs. The required training time was reduced by the order of 10e7. Despite a rather small training data set consisting of only 5 models, proliferation under drug treatment was well predicted. As a measure for prediction accuracy, the area under receiver operating characteristic curve (ROC AUC) was determined to be ~ 0.8, which I found pleasantly high. The source code for the modeling framework is hosted on GitHub and available at Benefit from such tools will be two-fold:

  1. Integration of genomic data harnessing ubiquitous sequencing efforts.
  2. Predictive models for the correct drug administration to inhibit tumor formation.

Unfortunately, I missed the second part of the session and the closing statement because I had to catch a train. Nevertheless, it became clear to me that the model-based discovery of drug-action mechanisms is already quite an achievement but not enough. If we all work together, we could advance systems biology of mammalian cells to the next level, creating a version 2.0, aka systems medicine. This approach, no matter how you call it, harbors the potential to aid people from physicians to patients by predicting specific treatment and guiding through personalized therapy. Unleashing this potential is our mission.




Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s