Patterns of time since last meal revealed by sparse PCA in an observational LC–MS based metabolomics study

TY - JOUR

T1 - Patterns of time since last meal revealed by sparse PCA in an observational LC–MS based metabolomics study

AU - Gürdeniz, Gözde

AU - Hansen, Louise

AU - Rasmussen, Morten Arendt

AU - Evrim, Acar Ataman

AU - Olsen, Anja

AU - christensen, Jane

AU - Barri, Thaer

AU - Tjønneland, Anne

AU - Dragsted, Lars Ove

N1 - CURIS 2013 NEXS 234

PY - 2013/10

Y1 - 2013/10

N2 - In metabolomics studies, liquid chromatography mass spectrometry (LC-MS) provides comprehensive information on biological samples. However, extraction of few relevant metabolites from this large and complex data is cumbersome. To resolve this issue, we have employed sparse principal component analysis (SPCA) to capture the underlying patterns and select relevant metabolites from LC-MS plasma profiles. The study involves a small pilot cohort with 270 subjects where each subject's time since last meal (TSLM) has been recorded prior to plasma sampling. Our results have demonstrated that both PCA and SPCA can capture the TSLM patterns. Nevertheless, SPCA provides more easily interpretable loadings in terms of selection of relevant metabolites, which are identified as amino acids and lyso-lipids. This study demonstrates the utility of SPCA as a pattern recognition and variable selection tool in metabolomics. Furthermore, amino acids and lyso-lipids are determined as dominating compounds in response to TSLM.

AB - In metabolomics studies, liquid chromatography mass spectrometry (LC-MS) provides comprehensive information on biological samples. However, extraction of few relevant metabolites from this large and complex data is cumbersome. To resolve this issue, we have employed sparse principal component analysis (SPCA) to capture the underlying patterns and select relevant metabolites from LC-MS plasma profiles. The study involves a small pilot cohort with 270 subjects where each subject's time since last meal (TSLM) has been recorded prior to plasma sampling. Our results have demonstrated that both PCA and SPCA can capture the TSLM patterns. Nevertheless, SPCA provides more easily interpretable loadings in terms of selection of relevant metabolites, which are identified as amino acids and lyso-lipids. This study demonstrates the utility of SPCA as a pattern recognition and variable selection tool in metabolomics. Furthermore, amino acids and lyso-lipids are determined as dominating compounds in response to TSLM.

U2 - 10.1007/s11306-013-0525-x

DO - 10.1007/s11306-013-0525-x

M3 - Journal article

SN - 1573-3882

VL - 9

SP - 1073

EP - 1081

JO - Metabolomics

JF - Metabolomics

IS - 5

ER -