Graphics / visualization challenges

[maxentile]

How do we visually summarize a MSM?

• Naive: draw a network where the nodes are exemplars of each state and edges are transition rates, e.g. (http://biomedicalcomputationreview.org/sites/default/files/u6/c_ntl9_jacs_fig3.jpg) (http://upload.wikimedia.org/wikipedia/commons/thumb/b/b9/ACBP_MSM_from_Folding@home.tiff/lossless-page1-350px-ACBP_MSM_from_Folding@home.tiff.png) ... Potentially with embellishments, e.g. a "potential energy surface" (http://portfolio.scistyle.com/Protein-Folding-Funnel)
  • Benefits: direct mapping to model representation in the computer
  • Limitations:
  • State markers: a state definition isn't just a single conformation, but a group of conformations. An individual conformation is difficult to interpret using a single 2D projection
  • Transition rates: we'll experience occlusion from many overlapping edges unless we arbitrarily threshold / sparsify the transition matrix
  • Doesn't "look dynamic"
  • Requires an additional marker (e.g. node outline color) to indicate the free energy of each conformation
  • Propagating probability mass multiple time steps ahead is difficult to do visually. If I start at a node, I look for and follow the couple biggest outgoing arrows and say most of the probability mass goes to those neighbors in one time step. I do the same thing for each of those nodes to figure out where the probability mass goes in two time steps. Etc. --> It would be cool to have an interface that automatically does this propagation for you. E.g. hover over a state, and then it does a looping animation where each frame indicates how much probability mass is on each node at a given time lag.

[maxentile]

Idea: interactive visualization. If we have an invertible mapping between reaction coordinates and conformations, then we could paint the free energy landscape as a heatmap and have the user drag the protein around on it. Each 2D position would correspond to a particular conformation. You could throw in additional embellishments, like having the location of the protein lag behind the location of the user's cursor as if connected by a spring, whose spring constant increases when crossing over free energy barriers. Without user input, the protein jiggles around in a random walk on this energy landscape. The markov state model definition would then be depicted as state boundaries on this landscape, with transition rates measured by counting the number of crossings observed.

As a learning tool, this could be very useful for introducing this way of thinking about things.

We might start with systems that are well-described by 2 natural collective variables, e.g. the alanine dipeptide. This then makes it clear why we care about finding good collective variables for high-dimensional systems.

[maxentile]

How can we visually summarize the results of molecular simulations? Inspecting molecular dynamics "movies" can be misleading. Also, what about MC simulations?

[maxentile]

What would be a fundamental enough problem to solve that I could submit to SIGGRAPH in January?

[maxentile]

Prior art: MSMexplorer:

• Bioinformatics paper: http://bioinformatics.oxfordjournals.org/content/29/7/950.long
• Github: https://github.com/brycecr/msmexplorer

Describes example use-case: discovery by Greg Bowman of "kinetic hubs" in protein folding pathways...

[maxentile]

Prior art: http://www.nature.com/srep/2014/140902/srep06264/full/srep06264.html -- Oh goodness, that first sentence, what am i getting myself into: "Present day science or more broadly society record observations as a function of time in diverse contexts[1]"