|
Activity Number:
|
318
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, August 2, 2016 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Statistics in Genomics and Genetics
|
|
Abstract #321250
|
View Presentation
|
|
Title:
|
Metabolomic Data Analysis: Data Preprocessing
|
|
Author(s):
|
Lauren McIntyre* and Rainey Patterson and Timothy Garrett and Alison Morse and Alexander Kirpich and Justin Fear and Miguel Ibarra and Oleksandr Moskalenko and Jeremy Koelmel
|
|
Companies:
|
University of Florida and University of Florida and University of Florida and University of Florida and University of Florida and National Institutes of Health and University of Florida and University of Florida and University of Florida
|
|
Keywords:
|
metabolomics ;
quality control ;
Galaxy
|
|
Abstract:
|
There is increasing interest in high throughput untargeted metabolic profiling. Yet, the typical data curation process is highly dependent upon expert judgement. Evaluation of a myriad of subjective choices for peak picking and detection thresholds has been similarly dependent on expert judgement. Here we develop several automated procedures for evaluation of high throughput untargeted LC-MS and apply these approaches to several datasets. Using blanks, noise can be significantly reduced with very small changes to false negative rates and a dramatic improvement to false positive rates. While peak picking, remains a difficult problem, measures of peak height are found to be less sensitive to variation in parameter choice during peak picking than peak area. A set of tools to facilitate comparisons are developed in python and also made available through the Galaxy interface (galaxyproject.org).
|
Authors who are presenting talks have a * after their name.
