Posters
« Back
PredRet: Prediction of Retention Time by Direct Mapping between Multiple Chromatographic Systems
EP23173
Poster Title: PredRet: Prediction of Retention Time by Direct Mapping between Multiple Chromatographic Systems
Submitted on 23 Jun 2015
Author(s): Jan Stanstrup1, Steffen Neumann2, Urška Vrhovšek1
Affiliations: 1Department of Food Quality and Nutrition, Research and Innovation Centre, Fondazione Edmund Mach, San Michele all’Adige (TN), Italy; 2Department of Stress and Developmental Biology, Leibniz Institute of Plant Biochemistry, Halle, Germany.
Poster Views: 1,314
View poster »


Poster Information
Abstract: Introduction
Increasingly sophisticated and more automated approaches to interpretation of mass spectra have long been the cornerstone of LC-MS based compound identification. Therefore the construction of MS databases have received considerable attention in the metabolomics community. While using fragmentation for compound identification is a powerful tool, it disregards fully half of the information provided by LC-MS.
For LC there are currently no coordinated efforts to share and exploit retention time (RT) information. RT information has been neglected in LC because the RT is specific to the chromatographic setup and there are no established RT references. A database of compounds’ RTs was therefore constructed and used to predict the RT of compounds in systems where the RTs had not been experimentally determined.

Methods
RTs of a number of compounds were experimentally determined in two different systems, and used to build monotonically increasing smooth generalized additive models between the RTs in the two systems. Building these models between all chromatographic systems in the database allowed the prediction of the RT for a high number of compounds in systems where they had not been experimentally determined.
The prediction tool has been made available as a web application at www.predret.org. The user can upload a spreadsheet with RTs of compounds measured in their system along with molecular identifiers such as PubChem CIDs or InChIs. This data is added to the database, models are calculated and predictions presented to the user.

Results
The initial database of RTs for > 2000 compounds was built from several in-house databases containing several hundred compounds each, 9 datasets available on MetaboLights containing a total of 640 compounds and the extensive dataset published by the Metabolic Systems Research Team at RIKEN covering 360 compounds.
An easy to use web interface, available at www.predret.org, was built for uploading experimental RTs and downloading RTs as predicted for the user’s own system.
Building models between all chromatographic systems in the database allowed for prediction of the RT of compounds in systems where they had not been experimentally determined. The number of compounds for which RTs can be predicted and the accuracy of the predictions is dependent on the number of compounds measured in both systems used in the mapping. With the current small database it was possible to predict up to 500 RTs with a median error between 0.02 and 0.25 min depending on the system. The median width of the confidence interval for each prediction currently ranges between 0.1 and 1.7 min.
The database can also be used to pinpoint likely erroneous user-reported RTs by comparison with predictions made from the models. We found that the majority of datasets contained entries with either misidentified compounds or incorrectly reported RT.
We believe that this tool will greatly help the identification process since compounds that are not compatible with the observed RT can be disregarded. Confirmatory experiments can then be reserved for compounds that could have the observed RT. This will allow researchers to complete the feature annotation and compound identification process in a faster and more rational manner and thus save time and resources, both monetary and environmental.
Community support is required to expand the database and thus dramatically improve the accuracy and coverage of the RT predictions.
Summary: Prediction of Retention Time by Direct Mapping between Multiple Chromatographic SystemsReport abuse »
Questions
Ask the author a question about this poster.
Ask a Question »

Creative Commons

Related Posters


Accelerated Ageing in Depression: A Study of Two Cohorts
Mathew A. Harris, Laura de Nooij, Xueyi Shen, Toni-Kim Clarke, Riccardo Marioni, Simon R. Cox, Emma L. Hawkins, Mark J. Adams, Liana Romaniuk, Stephen M. Lawrie, James H. Cole, Andrew M. McIntosh and Heather C. Whalley

Genetic Engineering in Male Sterility for Hybrid Variety Development
Abir Hasan Joy

Digiceuticals
Helana Lutfi and Shaban Nuredini

Applying Trapped Ion Mobility Separation (TIMS) in combination with Parallel Accumulation Serial Fragmentation (PASEF) for analysis of lipidomics samples
Sebastian Götz1, Sven W Meyer1,Ulrike Schweiger-Hufnagel1,Aiko Barsch1, Ningombam Sanjib Meitei2

BAT Molecular Imaging with SPECT-CT, PET-CT, PET-MRI and Fluorescence-PET: A Systematic Review of the Literature Data
Tarik Z Belhocine, MD., Ph.D *, Albert A Driedger, MD., Ph.D **, Jean-Luc Urbain, MD., Ph.D ***