Limits...
Piecewise multivariate modelling of sequential metabolic profiling data.

Rantalainen M, Cloarec O, Ebbels TM, Lundstedt T, Nicholson JK, Holmes E, Trygg J - BMC Bioinformatics (2008)

Bottom Line: We demonstrate the method on both simulated and metabolic profiling data, illustrating how time related changes are successfully modelled and predicted.The proposed method is effective for modelling and prediction of short and multivariate time series data.The method provides a competitive complement to commonly applied multivariate methods such as OPLS and Principal Component Analysis (PCA) for modelling and analysis of short time-series data.

View Article: PubMed Central - HTML - PubMed

Affiliation: Research Group for Chemometrics, Institute of Chemistry, Umeå University, Umeå, S-901 87, Sweden. mattias.rantalainen@imperial.ac.uk

ABSTRACT

Background: Modelling the time-related behaviour of biological systems is essential for understanding their dynamic responses to perturbations. In metabolic profiling studies, the sampling rate and number of sampling points are often restricted due to experimental and biological constraints.

Results: A supervised multivariate modelling approach with the objective to model the time-related variation in the data for short and sparsely sampled time-series is described. A set of piecewise Orthogonal Projections to Latent Structures (OPLS) models are estimated, describing changes between successive time points. The individual OPLS models are linear, but the piecewise combination of several models accommodates modelling and prediction of changes which are non-linear with respect to the time course. We demonstrate the method on both simulated and metabolic profiling data, illustrating how time related changes are successfully modelled and predicted.

Conclusion: The proposed method is effective for modelling and prediction of short and multivariate time series data. A key advantage of the method is model transparency, allowing easy interpretation of time-related variation in the data. The method provides a competitive complement to commonly applied multivariate methods such as OPLS and Principal Component Analysis (PCA) for modelling and analysis of short time-series data.

Show MeSH
Typical integrated 1H NMR spectrum from the HgCl2 data set.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC2373572&req=5

Figure 5: Typical integrated 1H NMR spectrum from the HgCl2 data set.

Mentions: To test the method on real data, we used data from a renal toxicity study using mercury II chloride to induce a proximal tubular damage [38] in the rat. This is a 1H NMR based metabonomic study of rat urine with data from seven time points (pre-dose, 0 h, 8 h, 24 h, 48 h, 72 h, 96 h) and including ten animals in total. Prior to analysis the data were pre-processed using standard methods. First the spectra were interpolated to a common chemical shift scale using cubic spline interpolation. The region corresponding to water and urea resonances (δ 4.5 – 6) was excluded from each spectrum and the spectral intensity was subsequently integrated over adjacent δ 0.04 ppm width bins. Each spectrum was normalized to the total sum of 100 units to reduce the overall dilution effect due to inter animal variability in urine excretion rates. A typical integrated NMR spectrum after pre-processing is shown in Figure 5. After 48 hours five animals were sacrificed, rendering N = 5 animals to be left in the study after 48 h.


Piecewise multivariate modelling of sequential metabolic profiling data.

Rantalainen M, Cloarec O, Ebbels TM, Lundstedt T, Nicholson JK, Holmes E, Trygg J - BMC Bioinformatics (2008)

Typical integrated 1H NMR spectrum from the HgCl2 data set.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC2373572&req=5

Figure 5: Typical integrated 1H NMR spectrum from the HgCl2 data set.
Mentions: To test the method on real data, we used data from a renal toxicity study using mercury II chloride to induce a proximal tubular damage [38] in the rat. This is a 1H NMR based metabonomic study of rat urine with data from seven time points (pre-dose, 0 h, 8 h, 24 h, 48 h, 72 h, 96 h) and including ten animals in total. Prior to analysis the data were pre-processed using standard methods. First the spectra were interpolated to a common chemical shift scale using cubic spline interpolation. The region corresponding to water and urea resonances (δ 4.5 – 6) was excluded from each spectrum and the spectral intensity was subsequently integrated over adjacent δ 0.04 ppm width bins. Each spectrum was normalized to the total sum of 100 units to reduce the overall dilution effect due to inter animal variability in urine excretion rates. A typical integrated NMR spectrum after pre-processing is shown in Figure 5. After 48 hours five animals were sacrificed, rendering N = 5 animals to be left in the study after 48 h.

Bottom Line: We demonstrate the method on both simulated and metabolic profiling data, illustrating how time related changes are successfully modelled and predicted.The proposed method is effective for modelling and prediction of short and multivariate time series data.The method provides a competitive complement to commonly applied multivariate methods such as OPLS and Principal Component Analysis (PCA) for modelling and analysis of short time-series data.

View Article: PubMed Central - HTML - PubMed

Affiliation: Research Group for Chemometrics, Institute of Chemistry, Umeå University, Umeå, S-901 87, Sweden. mattias.rantalainen@imperial.ac.uk

ABSTRACT

Background: Modelling the time-related behaviour of biological systems is essential for understanding their dynamic responses to perturbations. In metabolic profiling studies, the sampling rate and number of sampling points are often restricted due to experimental and biological constraints.

Results: A supervised multivariate modelling approach with the objective to model the time-related variation in the data for short and sparsely sampled time-series is described. A set of piecewise Orthogonal Projections to Latent Structures (OPLS) models are estimated, describing changes between successive time points. The individual OPLS models are linear, but the piecewise combination of several models accommodates modelling and prediction of changes which are non-linear with respect to the time course. We demonstrate the method on both simulated and metabolic profiling data, illustrating how time related changes are successfully modelled and predicted.

Conclusion: The proposed method is effective for modelling and prediction of short and multivariate time series data. A key advantage of the method is model transparency, allowing easy interpretation of time-related variation in the data. The method provides a competitive complement to commonly applied multivariate methods such as OPLS and Principal Component Analysis (PCA) for modelling and analysis of short time-series data.

Show MeSH