iPG2P Steering Committee Minutes
February 23, 2010; 1 to 2 pm CST
Present: Chris Myers, Doreen Ware (Christos, Jerry), Jeff White, Matt Vaughn, Ruth Grene, Steve Goff, Steve Welch, Tom Brutnell, Steve Rounsley, Dan Stanzione, Ed Buckler, Karla Gendler
The meeting was convened at 1pm CST and there was some general discussion to allow time for late arrivals to dial in.
Item 1: Review Action Items
Item 2: CI Development Update
Vaughn reported that not much user facing has changed in regards to CI Development. The provenance system and user authentication and authoring are being worked on now. For the next two-week cycle, interactions between the portal and back end services will be addressed along with provenance (once receive data model from Sudha Ram), continuation of moving prototype to production environment, and integrating security into services and the DE UI. UHTS is continuing detailed software requirements and Buckler wanted to make sure that variable length bar codes will be supported (they will be!). Stanzione added that there are only two formal releases scheduled, the iPToL release in March and the UHTS in June. With the Viz group, we will move into a prototyping phase that will start in the next two months but nothing at the production level.
Item 3: Facilitating GLM/GPU communication - Steve G/Steve W/Matt V
Goff reported that in response to Dave Lewenthall, John Hartman and Ali Akoglu, he and Martha Narro went to visit them to explain the basic biology behind QTL mapping. He and Narro were asked questions about datasets that the GPU had seen but were not able to answer questions relating to them (i.e., is this single regression? what are the rows and columns). It seems as though TASSEL is not transparent enough to understand QTL mapping and Goff came to the conclusion that there is a need to talk to the GPU group more and explain the biological problem and then ask if GPU is even the right implementation for the problem.
Welch added that he and Vaughn have been thinking about it and have been wondering if there was a need to look at another way to facilitate communication between the two groups, asked what failed in the first place (was it the biology explanation? Was it regression? Perhaps the initial problem wasn’t well defined). They began to toy with the idea of a F2F meeting with a small number of key people but wanted SC’s input on it.
Buckler stated that TASSEL was not written for NAM specifically but it is also not a public library. He thinks that a card-carrying mathematician/statistician is what is missing from the group. There are far too many biologists and the group needs someone the can bridge linear algebra and biology to help communicate. Goff has talked to Rebecca Doerge about the situation and now that there is a defined problem, she may be willing to help. Brutnell suggested Dan Nettleton. Buckler added that needed is someone who is good at doing sweeps in linear algebra. In response to Stanzione’s questions if the algorithms need improvement or if it is just necessary to explain the concept, Welch stated that they group is not asking for new algorithms but needs someone who can explain it and can converse at computation level and who also knows that GPU side of things. Buckler added that having Jean-Luc and Peter Bradbury working with that person would help. Welch will follow up with Doerge and then perhaps a meeting should be arranged between Doerge, Nettleton, Bradbury and the GLMxGPU group.
Item 4: Discussion
The meeting was opened for discussion. Grene said that if the ASPB proposal is accepted, at least one plant biologist should be there (Vaughn will be there to represent the plant biologists along with Steve Goff). Welch reported that Ann Stapleton has left Kansas State but is continuing to see if Arabidopsis flowering time model can represent wheat flowering. After the SC meeting in San Diego, White has pursued some modeling activities in maize working with Buckler’s group and has got interesting and mixed results. The process was very educational and White came away with the feeling that perhaps the model does not work well with environment conditions that were used. He would like to carry it forward but has to figure out how to fit it in with other priorities. He now has ViVa from Bernice and it appears straightforward and easy to work with.
Item 5: Identify Action Items
*Next meeting time: March 9, 2010; 1pm CST
Adjournment: The meeting ended at 2pm CST