SC_20110111

iPG2P Steering Committee
January 11, 2011; 2 to 3 pm EST

Present:

Notes/Agenda:
During our postdoc meeting last week, we spent some time discussing lessons learned from the tool integration homework assignments. To summarize: 1) simple transactional applications are easiest to describe while R packages and some JAVA applications require special treatment; 2) Complex configuration (decision trees) models are required in some cases; and 3) Older applications (MUSCLE, ClustalW) can have odd configuration modes or are more like comprehensive software suites. Matt then gave an overview of JSON and the iPlant Integration Model showing his work with BOWTIE as an example. Moving forward, work will begin on the actual integration after the API documents are finished this week. We will contact the postdocs individually for followup. Notes and slides from this meeting can be found here: https://pods.iplantcollaborative.org/wiki/display/ipg2p/PD_20110104.

Over the last few weeks, there has been significant progress made on Pat Schnable’s collaboration project with iPlant: a test for pairwise interactions for gene expression levels. The core algorithm is done, scaled, and runs on Longhorn with 1000x speedups, thus taking a 300 year problem and making it run in 4.5 hours on 128 cores. A visualization application was created. A paper was submitted to HICOMB 2011. Annalysis of the results is ongoing in collaboration with the Schable and Nettleton groups.

iPlant will have a significant presence at PAG this year. We encourage all of you attending to listen to any of our talks or visit any of our posters. Monday evening at 6pm in Pacific Salon 3, we will have a dedicated iPlant session. Additional talks and posters can be found here: https://pods.iplantcollaborative.org/wiki/display/ipg2p/PAG

Next meeting time:

Adjournment: