Points discussed include: Presenting an OpenWorm journal club on t…(more)
Points discussed include: Presenting an OpenWorm journal club on the recent stochastic neuronal model - Education Committee Synchronize with the larger OpenWorm roadmap Take the start of the c302-Sibernetic bridge and make it runnable and improve correctness (sibernetic) Improving neuronal model (c302) Improving synapse model (c302) Improving understanding of the whole nervous system circuit (neuronal-analysis) Aiming to have the TU Wien team able to implement one new circuit in c302 by end of November 2016
In the past releases we created a database of information about the…(more)
In the past releases we created a database of information about the c. elegans. Separately from this we created scripts that read spreadsheets in order to produce model files like the NeuroML connectome. This story is to create a single code base that can both serve as a means for an ever improving repository of data, and an automated way to produce model files for the simulation based on that data. Some principles to follow: There should be a one stop shop for accessing the data that goes into the model Model files (e.g. NeuroML) should be able to be easily updated as new information / data are integrated Data that are incorporated into the model should have references back to the source of that data. Currently the candidate codebase for this pipeline is the new PyOpenWorm library. More information about the data representation project can be found in the documentation
As a scientist or developer, I want to be able to run a test suite …(more)
As a scientist or developer, I want to be able to run a test suite against the simulation that will show me how close the model is to real data. In order for a model to demonstrate scientific value, it has to make falsifiable predictions. The target data to be able to predict will be drawn from the WormBehavior database. This milestone will involve working with these data, creating a code base that can compare movement output from the simulation with ground truth from the database and produce an accuracy score. More information about this story can be found on its project page in the documentation This story breaks down the epic to predict behavior from the WormBehavior database
As a user, I want to see the proof of concept sibernetic worm in my…(more)
Currently the Sibernetic code base does not output any sensory feed…(more)
Currently the Sibernetic code base does not output any sensory feedback into neurons. However, there is a Python layer that implements muscle activations. This story is to enable mechanosensory feedback coming from the model into the Python layer so it can be integrated into a model of 302 neurons. This means when the skin of the worm body is touched or, more generally, deformed, a signal can be received into a neuron. More information about the Sibernetic code is found in the documentation
As a user, I want to be able to see a visualization of the proof of…(more)
As a user, I want to be able to see a visualization of the proof of concept worm wiggling in my web browser and be able to perturb it in a manner that causes the wiggling to change in a realistic manner. This milestone suggests interactivity via Geppetto. The kind of perturbation is not defined yet-- ideally we should aim for the simplest kind we can think of that gives the user an interface to make modifications.
This epic is to have a simulation that can demonstrate it can predi…(more)
This epic is to have a simulation that can demonstrate it can predict (and therefore reproduce) 80% of the data collected about the N2 worm in the WormBehavior database. This means building a training set and a test set that are kept separate from each other, using the training set to tune up the model, then generating predictions, and comparing them against the test set, and doing some cross-validation. This epic focuses on an output of simulation performance rather than the means of implementation, so any way to achieve this epic is welcome.