Alberto Giovanni Busetto

Approximate Bayesian Computation

Approximate Bayesian computation (ABC) constitutes a class of computational methods rooted in Bayesian statistics. In all model-based statistical inference, the likelihood function is of central importance, since it expresses the probability of the observed data under a particular statistical model, and thus quantifies the support data lend to particular values of parameters and to choices among different models. For simple models, an analytical formula for the likelihood function can typically be derived. However, for more complex models, an analytical formula might be elusive or the likelihood function might be computationally very costly to evaluate. ABC methods bypass the evaluation of the likelihood function. In this way, ABC methods widen the realm of models for which statistical inference can be considered. ABC methods are mathematically well-founded, but they inevitably make assumptions and approximations whose impact needs to be carefully assessed. Furthermore, the wider application domain of ABC exacerbates the challenges of parameter estimation and model selection. ABC has rapidly gained popularity over the last years and in particular for the analysis of complex problems arising in biological sciences (e.g., in population genetics, ecology, epidemiology, and systems biology).

Unsupervised modeling of cell morphology dynamics for time-lapse microscopy

Analysis of cellular phenotypes in large imaging data sets conventionally involves supervised statistical methods, which require user-annotated training data. This paper introduces an unsupervised learning method, based on temporally constrained combinatorial clustering, for automatic prediction of cell morphology classes in time-resolved images. We applied the unsupervised method to diverse fluorescent markers and screening data and validated accurate classification of human cell phenotypes, demonstrating fully objective data labeling in image-based systems biology.

Multifaceted activities of type I interferon are revealed by a receptor antagonist.

A variant of type I interferon stimulates expression of only those genes required for an antiviral response. Building a Better Interferon Type I interferons stimulate an antiviral response during infections, inhibit cells from multiplying, and affect the immune system. The multiplicity of interferon actions complicates their use in patients. Interferons activate two distinct sets of genes: one for the protection against viruses and the other for the antiproliferative and immunomodulatory responses. Levin et al. found that the IFN-α2 variant interferon IFN-1ant prevented other forms of interferon from binding to the interferon receptor. However, at certain concentrations, IFN-1ant activated the genes required for antiviral immunity without activating the genes that suppress cell proliferation. Thus, IFN-1ant might be an effective therapy for treating viral infections. Type I interferons (IFNs), including various IFN-α isoforms and IFN-β, are a family of homologous, multifunctional cytokines. IFNs activate different cellular responses by binding to a common receptor that consists of two subunits, IFNAR1 and IFNAR2. In addition to stimulating antiviral responses, they also inhibit cell proliferation and modulate other immune responses. We characterized various IFNs, including a mutant IFN-α2 (IFN-1ant) that bound tightly to IFNAR2 but had markedly reduced binding to IFNAR1. Whereas IFN-1ant stimulated antiviral activity in a range of cell lines, it failed to elicit immunomodulatory and antiproliferative activities. The antiviral activities of the various IFNs tested depended on a set of IFN-sensitive genes (the “robust” genes) that were controlled by canonical IFN response elements and responded at low concentrations of IFNs. Conversely, these elements were not found in the promoters of genes required for the antiproliferative responses of IFNs (the “tunable” genes). The extent of expression of tunable genes was cell type–specific and correlated with the magnitude of the antiproliferative effects of the various IFNs. Although IFN-1ant induced the expression of robust genes similarly in five different cell lines, its antiviral activity was virus- and cell type–specific. Our findings suggest that IFN-1ant may be a therapeutic candidate for the treatment of specific viral infections without inducing the immunomodulatory and antiproliferative functions of wild-type IFN.

Automatic generation of predictive dynamic models reveals nuclear phosphorylation as the key Msn2 control mechanism.

Topological filtering identifies biological networks compatible with known data and enables quantitative analysis of regulatory mechanisms. Reducing the Options Quantitative analysis of signaling systems is challenging because limited quantitative data are available and the data can be represented by many network models. Sunnåker et al. developed a computational approach called topological filtering to systematically and automatically integrate modeling and data acquisition to infer the set of mechanistically plausible models, thus vastly reducing the number of potential models. The approach iteratively eliminates reactions from the model to identify only those topological networks that fit the data. Application of their method to an extracellular signal–regulated kinase (ERK) pathway that could be represented by 512 possible network topologies reduced the possibilities to 16 and showed that a set of feedback reactions were necessary to quantitatively represent the results. Topological filtering applied to the regulation of the localization of Msn2, a yeast transcription factor controlled by phosphorylation by PKA (protein kinase A) in response to changes in glucose abundance, identified a single model that fit the data. Comparison of model predictions with experimental data showed that the nuclear phosphorylation rate was key to controlling Msn2 nuclear abundance in response to cAMP (cyclic adenosine monophosphate), a signal produced as cells recover from glucose starvation. Predictive dynamical models are critical for the analysis of complex biological systems. However, methods to systematically develop and discriminate among systems biology models are still lacking. We describe a computational method that incorporates all hypothetical mechanisms about the architecture of a biological system into a single model and automatically generates a set of simpler models compatible with observational data. As a proof of principle, we analyzed the dynamic control of the transcription factor Msn2 in Saccharomyces cerevisiae, specifically the short-term mechanisms mediating the cells’ recovery after release from starvation stress. Our method determined that 12 of 192 possible models were compatible with available Msn2 localization data. Iterations between model predictions and rationally designed phosphoproteomics and imaging experiments identified a single-circuit topology with a relative probability of 99% among the 192 models. Model analysis revealed that the coupling of dynamic phenomena in Msn2 phosphorylation and transport could lead to efficient stress response signaling by establishing a rate-of-change sensor. Similar principles could apply to mammalian stress response pathways. Systematic construction of dynamic models may yield detailed insight into nonobvious molecular mechanisms.

Near-optimal experimental design for model selection in systems biology

Motivation: Biological systems are understood through iterations of modeling and experimentation. Not all experiments, however, are equally valuable for predictive modeling. This study introduces an efficient method for experimental design aimed at selecting dynamical models from data. Motivated by biological applications, the method enables the design of crucial experiments: it determines a highly informative selection of measurement readouts and time points. Results: We demonstrate formal guarantees of design efficiency on the basis of previous results. By reducing our task to the setting of graphical models, we prove that the method finds a near-optimal design selection with a polynomial number of evaluations. Moreover, the method exhibits the best polynomial-complexity constant approximation factor, unless P = NP. We measure the performance of the method in comparison with established alternatives, such as ensemble non-centrality, on example models of different complexity. Efficient design accelerates the loop between modeling and experimentation: it enables the inference of complex mechanisms, such as those controlling central metabolic operation. Availability: Toolbox ‘NearOED’ available with source code under GPL on the Machine Learning Open Source Software Web site (mloss.org). Contact: busettoa@inf.ethz.ch Supplementary information: Supplementary data are available at Bioinformatics online.

Inferring causal metabolic signals that regulate the dynamic TORC1-dependent transcriptome

Cells react to nutritional cues in changing environments via the integrated action of signaling, transcriptional, and metabolic networks. Mechanistic insight into signaling processes is often complicated because ubiquitous feedback loops obscure causal relationships. Consequently, the endogenous inputs of many nutrient signaling pathways remain unknown. Recent advances for system‐wide experimental data generation have facilitated the quantification of signaling systems, but the integration of multi‐level dynamic data remains challenging. Here, we co‐designed dynamic experiments and a probabilistic, model‐based method to infer causal relationships between metabolism, signaling, and gene regulation. We analyzed the dynamic regulation of nitrogen metabolism by the target of rapamycin complex 1 (TORC1) pathway in budding yeast. Dynamic transcriptomic, proteomic, and metabolomic measurements along shifts in nitrogen quality yielded a consistent dataset that demonstrated extensive re‐wiring of cellular networks during adaptation. Our inference method identified putative downstream targets of TORC1 and putative metabolic inputs of TORC1, including the hypothesized glutamine signal. The work provides a basis for further mechanistic studies of nitrogen metabolism and a general computational framework to study cellular processes.

Modelling and optimizing the process of learning mathematics

This paper introduces a computer-based training program for enhancing numerical cognition aimed at children with developmental dyscalculia. Through modelling cognitive processes and controlling the level of their stimulation, the system optimizes the learning process. Domain knowledge is represented with a dynamic Bayesian network on which the mechanism of automatic control operates. Accumulated knowledge is estimated to select informative tasks and to evaluate student actions. This adaptive training environment equally improves success and motivation. Large-scale experimental data quantifies substantial improvement and validates the advantages of the optimized training.

Optimized expected information gain for nonlinear dynamical systems

This paper addresses the problem of active model selection for nonlinear dynamical systems. We propose a novel learning approach that selects the most informative subset of time-dependent variables for the purpose of Bayesian model inference. The model selection criterion maximizes the expected Kullback-Leibler divergence between the prior and the posterior probabilities over the models. The proposed strategy generalizes the standard D-optimal design, which is obtained from a uniform prior with Gaussian noise. In addition, our approach allows us to determine an information halting criterion for model identification. We illustrate the benefits of our approach by differentiating between 18 published biochemical models of the TOR signaling pathway, a model selection problem in systems biology. By generating pivotal selection experiments, our strategy outperforms the standard Aoptimal, D-optimal and E-optimal sequential design techniques.

Stable Bayesian Parameter Estimation for Biological Dynamical Systems

The estimation of kinetic rate constants plays a key role for the development of dynamical models in systems biology. Bayesian inference addresses the issues of noise modelling and quantification of parameter uncertainty. However, current approximate inference techniques suffer from well-known degeneracy and instability problems. We propose a novel Bayesian inference technique to estimate parameters of biological dynamical systems in a convergent and stable way. Our approximation is based on sequential Monte Carlo resampling of belief states according to clusters of particles. The resulting implicit partitions of the parameter space keep the density of samples high in the most informative regions. The method yields two highly desirable results: sample degeneracy is avoided by preventive resampling, while modal instability is contrasted by particle clustering. We have tested our approach on the double Goodwin model. As we show, our strategy improves the stability compared to current methods: at the same computational cost, it is successful in maintaining the required modes where standard approaches systematically fail. Moreover, our strategy suggests regions of interest in the parameter space which cannot be identified by traditional resampling schemes. We expect such improvements to open the way for a better understanding of the dynamical behaviors of nonlinear systems in computational science and engineering.

Cluster-Based Prediction of Mathematical Learning Patterns

This paper introduces a method to predict and analyse students’ mathematical performance by detecting distinguishable subgroups of children who share similar learning patterns. We employ pairwise clustering to analyse a comprehensive dataset of user interactions obtained from a computer-based training system. The available data consist of multiple learning trajectories measured from children with developmental dyscalculia, as well as from control children. Our online classification algorithm allows accurate assignment of children to clusters early in the training, enabling prediction of learning characteristics. The included results demonstrate the high predictive power of assignments of children to subgroups, and the significant improvement in prediction accuracy for short- and long-term performance, knowledge gaps, overall training achievements, and scores of further external assessments.