Surprisal analysis (SA) is a (bio)molecule-centric thermodynamic approach grounded in information theory. When applied to genomic and metabolic data, SA allows understanding and predicting biological changes of phenotype induced by perturbations.<br/>SA allows determining a balanced state that is common to all the samples and whose phenotype gathers the balanced cellular processes, such as ‘house keeping’ genes that maintain basic cellular functions and ensure the cell stability. This balanced state was shown to be common to several types of cellular systems. The balanced state determines the thermodynamic maximal entropy that the system can reach. The actual entropy of the system can be lower or equal to that determined by the balance state because the cellular system is subject to environmental and genomic constraints which represent the unbalanced cellular processes as sampled by the data set. SA allows quantifying the importance of these constraints and interpreting the gene or metabolite that they determine in terms of a set of gene networks involved in the unbalanced processes.<br/>To summarize, using SA, the free energy of each gene or metabolite level is compacted as the sum of a term corresponding to the balance state common to all the samples present in the data set plus a few additional terms that quantify the effect of the constraints common to subsets of samples. For example, the figure below shows the weight of the balanced state, λ<sub>0</sub>, is common for all the 20 samples of a data set comprising control and patient diagnosed with GBM cancer, whether they are control or diseased. The first constraint sorts between the control from the diseased patients while subsequent constraints identify diseased specific patterns. We will discuss the methodology and its application to two types of biological samples: cancer patient genomic data and unicellular algae genomic and metabolic data.

Thermodynamic-like information theoretic surprisal analysis of genomic and metabolic data

256th National Meeting (2018)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.