iPToL_ET_24JUN09

Owned by Sheldon McKay
Last updated: Sept 08, 2009Version comment
3 min read

Attendees: Sriram, Andy, Nick, Nirav, Karla, Matt, Adam, Damian, Jerry, Liya, Scott, Sheldon, Dan,

Action Items:

  • Follow through on web 2 evaluations.
  • Karla and Sheldon will poke data assembly, data recognition, and tree reconciliation.

Agenda

  • Review 10JUN09 Action Items (Sheldon)
  • Tool Reports (Scott)
  • General Update (Sheldon)
  • Data Integration Update (Sheldon)
  • Trait Mapping Update (Karla)
  • Big Trees Update (Nirav)
  • Open Discussion

Notes

General Update (Sheldon):

Karla and Sheldon have been arranging meeting with working group leads. Brian Omerara for trait recognition working groups. Val Tannen/Bill Piel meeting didn’t meet needs to get specific guidance on direction. ToL proposal ($1.9) proposes to do many of the same things in computer visualization that we’ve agreed to do. We will have to address visualization needs in the shorter term. Results from Nescent/phylogeny hack-a-thon in March: 1) data interoperability project 2) a new group that focuses on interoperability. Next week, Todd Vision (or designate) will present next steps in the tree reconciliation group. Currently, agreement is being circulated and commented on within iPToL group.

Trait Mapping Update:

Goals are right in line with the DE, receives trees from user, analyze the tree, report results getting tree displays. Needs to provide portals to larger databases. The heavy lifting will be setting up and connecting the analytical pipelines. Independent contrasts should be the first thing we tackle. Top goal - A web app that is used to do analyze quickly and easily. Moderate goal - shows contrasts. Users: build trees for a living and 2) don’t care just use trees. Timeline: maybe taking full two years if the apps can scale up. We will have to start testing programs to see where they fail. Brian will create a 50K species test tree with 2 continuous strains and 2 discreet strains. We might have to retool algorithms to optimize. Optimizing means speeding them up. They run slowly with large data sets. Large is 5K taking 2-4 hours. Karla and Sheldon should come up with an evaluation matrix to test the algorithms.

Big Trees Update:

We are investigating checkpointing for lightly virtualization layers. Sand Diego group has taken on looking at the code on the threading and paralleling aspects. CP w/o toughing the code, looking at 4 different facilities to do that. Some notion of a pipeline and benchmarks of how scalable it was, where it fell off the wagon, etc. There may be some convergence in activities with Alexis and Brian. We will run a 15 min test and then rerun to see if we get the same results – same results would be success. Wayne P has access to Intel multicore processors and is cleaning up the code on the surface.

Open Discussion:

Sheldon gave top at Evolution meeting giving a general overview of ToL working groups and engagement team. There was no time for questions for no real feedback. Evaluating text book called “Phylogenetic Trees Made Easy” (Gary Hall). Will give update to see if it will work as a primer.

There is information under the links sections and in the document libraries.

We should evolve an articulated policy with our communication tools.
Alfresco - doc mgmt system
Confluence – wiki

iPlant-wide Talk repository in Alfresco to archive talks.

No Wikipedia page for iPlant. – waiting on someone else to create it so that we can then add content.

Pam Soltis, Edu director of ToL. Asked for a more detailed proposal for but didn’t’ get a great response. One of the problems of the evaluation is that the pipeline was broken. Cam Webb restored their workflow and forward a doc that is more detailed than the original email.