SURAgrid Application Description

Application Name
/ Multiple Imputation on Genetics (MIG) Analysis
Application Area / Bio-Statistics, research on diabetes
Keywords / R-Statistical language, embarrassingly parallel, inputs
Project/Dept. Affiliation / UAB Biostatistics, Section on Statistical Genetics (SSG) in collaboration with UAB HPC Services Group, UAB
Specific Objective for Using SURAgrid / To add resources to UAB compute pool and thereby increase the
speed and throughput of this computations.
To explore the use of GridWay meta-scheduling on SURAgrid.
To document the process/experience of resource acquisition via SURAgrid.
Originating institution / University of Alabama at Birmingham (UAB), Birmingham, AL
Primary Application Contact Name, Email / Poornima Pochana (), John-Paul Robinson ()
Technical Contact:
Name, Email / Poornima Pochana ()
Participating sites
Application Description / The Section on Statistical Genetics (SSG) of UAB applies
statistical genetics methods to understand gene expression and
how it relates to expressed traits. They make use of R-statistical
language to generate the statistics.

Multiple Imputations is a kind of statistical method used for

analyzing gene expression. SSG of UAB performs MIG analysis

with the help of methods written in R – statistical language.

Do you have HPC available at your institution? / Yes
Are you interested in contributing a resource to SURAgrid? / UAB is already a registered contributor to SURAgrid
Initial (Test) / Future (Production)
What are the anticipated system requirements for SURAgrid nodes running this application? / Globus, R-statistical language tool of version >= 2.6.1 (R optional if user account allows code installs)
Have you attended any HPC or Grid training? If so, which ones? / Yes (training as HPC employee of UAB)
Grid focus (data sharing, computation, access to unique resources, collaboration) / computation, collaboration
Does your application have any network dependencies (bandwidth, latency, multicast, other)? / no
What is the expected frequency of application run (one-time, occasional, monthly, weekly, daily…)? / daily during devel, less frequently in production.
Describe expected application invocation mechanism (by user submitting job, programmatically by some event or timing…) /
  1. Through direct submission on the resource
qsub -t 1-100 job.sh 10000 100 mig.txt
  1. Through Gridway meta-scheduler
gwsubmit -v -t job.jt
Estimated start date for application run / 09/12/2008
Is this application open to others to use with their own data or revisions? / Yes
Do you have an access to a Certificate Authority?
If yes, location is your Certificate Policy and Certificate Practice? Contact for your CA? / Yes.


John-Paul Robinson,
Are you targeting any particular SURAgrid resource? Why? / No
Additional comments

<template rev. 2: August 2007>1