Feed-forward motifs in transcription factor networks evolved to filter out spurious signals? ‘Just-so’ no longer

Biosystems Analytics

Mechanistic computational models, particularly rule-based stochastic models, are a vital complement to wet-lab experiments (and a vital chunk of our work at Amber Biology), but can also provide insights into evolutionary processes. In a paper just published in Nature Communications, the team, which included Kun Xiong, myself, Mark Siegal and Joanna Masel, asked whether a particular 3-node feed-forward loop motif (specifically the type 1 coherent FFL, or C1-FFL, widely hypothesized to have evolved to filter out spurious signals, actually evolved for that purpose. Due to it’s overrepresentation in the transcriptional networks of many species, and it’s demonstrated function in filtering out these signals many researchers have previously accepted a kind of ‘just-so’ account of the feed-forward motif. To test this hypothesis properly, we built a detailed stochastic model of the dynamics of transcriptional networks, and then allowed the network to evolve under selection for the function, and without…

View original post 103 more words

Advertisements

The limitations of Big Data in life science R&D

Biosystems Analytics

Big Data has become an increasingly large presence in the life science R&D world, but as I have blogged about previously, increasingly larger datasets and better machine algorithms alone, will not leverage that data into bankable knowledge and can lead to erroneous inferences.  My Amber Biology colleague, Gordon Webster has a great post over on LinkedIn leavening the hype around Big Data, pointing out that analytics and visualizations alone are insufficient for making progress in extracting knowledge from biological datasets:

Applying the standard pantheon of data analytics and data visualization techniques to large biological datasets, and expecting to draw some meaningful biological insight from this approach, is like expecting to learn about the life of an Egyptian pharaoh by excavating his tomb with a bulldozer

“-omics” such as those produced by transcriptomic and proteomic analyses are ultimately generated by dynamic processes consisting of individual genes, proteins and other molecules…

View original post 130 more words

Connecting the cognitive dots in systems biology modeling

Biosystems Analytics

Building computational models in any discipline has many challenges starting at inclusion (what goes in, what’s left out), through to representation (are we keeping track of aggregate numbers, or actual individuals), implementation (efficiency, cost) and finally  verification and validation (is it correct?).  Creating entire modeling softwareplatforms intended for end-user scientists within a discipline brings an entirely new level of challenge.  Cognitive issues of representation within the modeling platform – always present when trying to communicate the content of a model to others – become one of the most central challenges.  To create modeling platforms that, say, a biologist might want to use, requires paying close attention to the idioms and metaphors used at the most granular level of biology: at the whiteboard, the bench, or even in the field.

Constructing such software with appropriate metaphors, visual or otherwise, requires close collaboration with working scientists at every…

View original post 402 more words