...
Recap of Day 1
Welch started the meeting by recapping yesterday's yesterday’s talks and decisions. The group talked about kinds of things they wanted to see; an initial release focusing on parameter estimation and sensitivity analysis; links to visualization programs; specific details of algorithms used and exports that they'd they’d like to see from the system.
...
There are two types of metadata: experimental or descriptive metadata and a semantic description of what the data is. Descriptive is experimental data and/or computational provenance within the domain of the system; semantic describes what it is (i.e. tabular data) and is used for at a higher level of control. One principle to follow would be that a user doesn't doesn’t lose any of where the data came from (for credit and publications purposes). iPlant does not want to destroy, discard, and/or modify any of the metadata coming in. With modeling, there is probably unique data that needs to be captured and the group should define what that is. It is safe to assume that iPlant will store all of the basic information. Welch stated that the group has two things to talk about: 1) specific data sets that we know are out there relating to use cases and 2) data items relating to metadata.
...
In regards to datasets, Myers is hoping to get his manuscript out so that the data will become available to the public. Stapleton suggested Mark Stitiz's Stitiz’s Arabidopsis data. Myers wondered how ready the model is off the shelf and Welch suggested that the group can look at it and decide. Vaughn asked how do you hook up the photosynthesis model physically. Both models are rate driven models. At the very simplest level, you would pick an output of the model and use it as a "sugar signal" “sugar signal” or proxy. In the phenology model, you would then use that variable to drive it. Myers added that there are specific data sets that are being generated and maybe an idea would be to look at genetic variation in Arabidopsis.
...
For major classes of ecophysiological models, there are standard formats developed. These could be used to get an idea and identify categories of information that we've we’ve talked about. Myers said that there is a need for data for the models but the group also should identify what needs to come out of the models. Welch suggested that light and temperature are in common across models, with enzyme levels and planting dates differing. Myers added that SBML is very good for capturing data but is not good at describing it. His group has had to think through a set of abstractions that describe the data. Vaughn said that a controlled vocabulary is needed. The suggestion was to have White send Chris Jordan his data file and ask him what would he do. White also presented some of his visualization work related to the NAM work and this can be found here .
Welch summarized that a need that the group sees at this point relative to modeling will be ways to manage data that is in the collection process (tools that make it easy to collect metadata/provenance that puts them in to tabular data model that Jordan had identified). Jordan and the group will want to look at White's White’s data scheme as a test case to identify issues when it comes be worked with in terms of encapsulation.
Prioritize EOT Aspects & Action Items
Vaughn began discussion by talking about points that were brought up over dinner last night. DNA Subway is an excellent example of an educational based DE and it would be nice to identify similar overlap and synergy with modeling. Some ideas would include taking the NAM data and projecting it across expected global climate change models where the models run in real time and one can slide forward in time or change parameters and see how model predictions change in response to global climate change. In reference to Ed's Ed’s project, it is nice but it would just be another avenue that iPlant would have to pursue. He would like to see educational requirements exist as first class citizens along with the scientific requirements. He suggested focusing on an application that runs a very specific model and then iPlant can make it slick.
Myers suggested that the group could exploit models with "what“what-if" if” scenarios. White added that class activities would have to be low time commitment and things that are science fair type of projects. Stapleton said that the buzz phrase is "authentic experience"“authentic experience”, someone other than you, the student, has to care about the work/experiment you are doing. Welch commented that the group could have workshops for teachers where some information on plants is presented with this information is assisted by models. Myers proposed using the programming skills of younger kids to help with open-source of DSSAT and to look into Google Summer of Code (maybe next year?). He also said that the climate change aspect of things could be a major hook.
...