Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
1/22 jrwb.de
Progress in regulatory degradation kineticsSETAC Europe 24th Annual Meeting
PD Dr. Johannes RankeScientific consultant
Basel, 13 May 2014
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
2/22 jrwb.de
How do you define progress?
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
3/22 jrwb.de
Outline
1 The tasks of regulatory degradation kinetics
2 Parameter confidence intervals
3 Biphasic models for metabolites
4 Conclusions
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
3/22 jrwb.de
Outline
1 The tasks of regulatory degradation kinetics
2 Parameter confidence intervals
3 Biphasic models for metabolites
4 Conclusions
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
3/22 jrwb.de
Outline
1 The tasks of regulatory degradation kinetics
2 Parameter confidence intervals
3 Biphasic models for metabolites
4 Conclusions
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
3/22 jrwb.de
Outline
1 The tasks of regulatory degradation kinetics
2 Parameter confidence intervals
3 Biphasic models for metabolites
4 Conclusions
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
4/22 jrwb.de
The tasks of regulatory degradation kinetics
Derive endpoints for fate modelling
Provide endpoints for comparison with trigger values
Triggers for further data requirements(EU pesticides: FOCUS kinetics “persistence endpoints”)Triggers for persistence, P and vP(Regulation 1107/2009, REACH)
Reflect endpoint uncertainty
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
4/22 jrwb.de
The tasks of regulatory degradation kinetics
Derive endpoints for fate modelling
Provide endpoints for comparison with trigger values
Triggers for further data requirements(EU pesticides: FOCUS kinetics “persistence endpoints”)Triggers for persistence, P and vP(Regulation 1107/2009, REACH)
Reflect endpoint uncertainty
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
4/22 jrwb.de
The tasks of regulatory degradation kinetics
Derive endpoints for fate modelling
Provide endpoints for comparison with trigger values
Triggers for further data requirements(EU pesticides: FOCUS kinetics “persistence endpoints”)
Triggers for persistence, P and vP(Regulation 1107/2009, REACH)
Reflect endpoint uncertainty
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
4/22 jrwb.de
The tasks of regulatory degradation kinetics
Derive endpoints for fate modelling
Provide endpoints for comparison with trigger values
Triggers for further data requirements(EU pesticides: FOCUS kinetics “persistence endpoints”)Triggers for persistence, P and vP(Regulation 1107/2009, REACH)
Reflect endpoint uncertainty
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
4/22 jrwb.de
The tasks of regulatory degradation kinetics
Derive endpoints for fate modelling
Provide endpoints for comparison with trigger values
Triggers for further data requirements(EU pesticides: FOCUS kinetics “persistence endpoints”)Triggers for persistence, P and vP(Regulation 1107/2009, REACH)
Reflect endpoint uncertainty
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
5/22 jrwb.de
Motivation to improve regulatory degradationkinetics
Provide the best possible foundation
Transparency
X
Scientific quality
?
Technical quality
Collaboration
!
www.r-project.org - focus.jrc.ec.europa.eu/dkR CMD check mkin 0.9-27.tar.gz - github.com/jranke/mkin
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
5/22 jrwb.de
Motivation to improve regulatory degradationkinetics
Provide the best possible foundation
Transparency
X
Scientific quality
?
Technical quality
Collaboration
!
www.r-project.org - focus.jrc.ec.europa.eu/dkR CMD check mkin 0.9-27.tar.gz - github.com/jranke/mkin
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
5/22 jrwb.de
Motivation to improve regulatory degradationkinetics
Provide the best possible foundation
Transparency
X
Scientific quality
?
Technical quality
Collaboration
!
www.r-project.org - focus.jrc.ec.europa.eu/dkR CMD check mkin 0.9-27.tar.gz - github.com/jranke/mkin
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
5/22 jrwb.de
Motivation to improve regulatory degradationkinetics
Provide the best possible foundation
Transparency
X
Scientific quality
?
Technical quality
Collaboration
!
www.r-project.org - focus.jrc.ec.europa.eu/dkR CMD check mkin 0.9-27.tar.gz - github.com/jranke/mkin
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
5/22 jrwb.de
Motivation to improve regulatory degradationkinetics
Provide the best possible foundation
Transparency X
Scientific quality
?
Technical quality
Collaboration
!
www.r-project.org
- focus.jrc.ec.europa.eu/dkR CMD check mkin 0.9-27.tar.gz - github.com/jranke/mkin
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
5/22 jrwb.de
Motivation to improve regulatory degradationkinetics
Provide the best possible foundation
Transparency X
Scientific quality ?
Technical quality
Collaboration
!
www.r-project.org - focus.jrc.ec.europa.eu/dk
R CMD check mkin 0.9-27.tar.gz - github.com/jranke/mkin
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
5/22 jrwb.de
Motivation to improve regulatory degradationkinetics
Provide the best possible foundation
Transparency X
Scientific quality ?
Technical quality
Collaboration
!
www.r-project.org - focus.jrc.ec.europa.eu/dkR CMD check mkin 0.9-27.tar.gz
- github.com/jranke/mkin
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
5/22 jrwb.de
Motivation to improve regulatory degradationkinetics
Provide the best possible foundation
Transparency X
Scientific quality ?
Technical quality
Collaboration !
www.r-project.org - focus.jrc.ec.europa.eu/dkR CMD check mkin 0.9-27.tar.gz - github.com/jranke/mkin
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
6/22 jrwb.de
Critical areas
t-test for parameter significance (assumes normaldistribution for estimator)
Confidence intervals for fitted parameters
Modelling biphasic behaviour of metabolites
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
7/22 jrwb.de
Elements of success
mkin was first published in May 2010, including biphasic models formetabolites since May 18. It was then used to develop
KinGUII (Bayer Crop Science)
CAKE (Syngenta)
by adding a graphical user interface (GUI), iteratively reweightedleast squares (IRLS) and Markov Chain Monte Carlo (MCMC)
Isometric logratio transformation (ILR) for fitting formation fractionstogether with Rene Lehmann (UBA) in 2012
Parameter confidence intervals based on transformed parameters(2013)
mkin (≥ 0.9-27) allows for fitting models with or without formationfractions, with or without parameter transformations.
Ranke and Lehmann, SETAC World 20-24 May 2012, Berlinmkin 0.9-27 published on CRAN 10 May 2014
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
7/22 jrwb.de
Elements of success
mkin was first published in May 2010, including biphasic models formetabolites since May 18. It was then used to develop
KinGUII (Bayer Crop Science)
CAKE (Syngenta)
by adding a graphical user interface (GUI), iteratively reweightedleast squares (IRLS) and Markov Chain Monte Carlo (MCMC)
Isometric logratio transformation (ILR) for fitting formation fractionstogether with Rene Lehmann (UBA) in 2012
Parameter confidence intervals based on transformed parameters(2013)
mkin (≥ 0.9-27) allows for fitting models with or without formationfractions, with or without parameter transformations.
Ranke and Lehmann, SETAC World 20-24 May 2012, Berlin
mkin 0.9-27 published on CRAN 10 May 2014
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
7/22 jrwb.de
Elements of success
mkin was first published in May 2010, including biphasic models formetabolites since May 18. It was then used to develop
KinGUII (Bayer Crop Science)
CAKE (Syngenta)
by adding a graphical user interface (GUI), iteratively reweightedleast squares (IRLS) and Markov Chain Monte Carlo (MCMC)
Isometric logratio transformation (ILR) for fitting formation fractionstogether with Rene Lehmann (UBA) in 2012
Parameter confidence intervals based on transformed parameters(2013)
mkin (≥ 0.9-27) allows for fitting models with or without formationfractions, with or without parameter transformations.
Ranke and Lehmann, SETAC World 20-24 May 2012, Berlin
mkin 0.9-27 published on CRAN 10 May 2014
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
7/22 jrwb.de
Elements of success
mkin was first published in May 2010, including biphasic models formetabolites since May 18. It was then used to develop
KinGUII (Bayer Crop Science)
CAKE (Syngenta)
by adding a graphical user interface (GUI), iteratively reweightedleast squares (IRLS) and Markov Chain Monte Carlo (MCMC)
Isometric logratio transformation (ILR) for fitting formation fractionstogether with Rene Lehmann (UBA) in 2012
Parameter confidence intervals based on transformed parameters(2013)
mkin (≥ 0.9-27) allows for fitting models with or without formationfractions, with or without parameter transformations.
Ranke and Lehmann, SETAC World 20-24 May 2012, Berlinmkin 0.9-27 published on CRAN 10 May 2014
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
8/22 jrwb.de
Soil metabolism of 2,4,5-T in the labCl
Cl
Cl
O O
O
0 50 100 150 200
020
4060
80Commerce
Time
Obs
erve
d
●
●
●
●
● ●● ●
0 50 100 150 200
020
4060
80
Catlin
Time
Obs
erve
d
●
●
●
●● ● ● ●
0 50 100 150
020
4060
80
Fargo
Time
Obs
erve
d
●●●
●
●
●
●
●
0 50 100 150
020
4060
80
Keith
Time
Obs
erve
d
●●●
●
●
● ● ●
0 50 100 150
020
4060
80
Walla Walla
Time
Obs
erve
d
●
●
●
●
●● ●
0 50 100 150
020
4060
80
Cecil
Time
Obs
erve
d
●
●
●
●
●●
●
● T245phenolanisole
McCall et al. (1981) J Agric Food Chem 29 100-107
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
9/22 jrwb.de
2,4,5-T in Commerce soil in gmkin
gmkin user interface
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
10/22 jrwb.de
Pathway from 2,4,5-T-phenol to sink?Cl
Cl Cl
OH
k phenol sink negligible - fix to zero
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
10/22 jrwb.de
Pathway from 2,4,5-T-phenol to sink?Cl
Cl Cl
OH
k phenol sink negligible - fix to zero
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
10/22 jrwb.de
Pathway from 2,4,5-T-phenol to sink?Cl
Cl Cl
OH
k phenol sink negligible - fix to zero
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
11/22 jrwb.de
2,4,5-T in Commerce soil, no path to sink
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
11/22 jrwb.de
2,4,5-T in Commerce soil, no path to sink
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
12/22 jrwb.de
Rate parameters log transformed during fit
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
13/22 jrwb.de
Proposal regarding t-test
Instead of testing rate constants for significant difference fromzero:
Use best available estimate
Consider if it is neglibly small
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
13/22 jrwb.de
Proposal regarding t-test
Instead of testing rate constants for significant difference fromzero:
Use best available estimate
Consider if it is neglibly small
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
14/22 jrwb.de
2,4,5-T in Fargo soil
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
15/22 jrwb.de
2,4,5-T in Fargo soilModel with formation fractions
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
16/22 jrwb.de
2,4,5-T in Fargo soilModel with transformed formation fractions
We get a plausible confidence interval without doing an MCMC
simulation
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
17/22 jrwb.de
Parallel formation of metabolites
Confidence intervals only for single formation fractions(slide corrected after the meeting)
Data from Schafer et al. (2007) Proc. XIII SymposiumPesticide Chemistry, Piacenza, 2007, p. 916-923
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
18/22 jrwb.de
Conceptual comparison of DFOP and SFORB
Both possible for metabolites, but with two extra parameters.
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
19/22 jrwb.de
Alternative with one extra parameter
We could also use an Indeterminate Order Rate Equation(IORE) for metabolites
dm
dt= ...− kmm
n...
This is used in North America for parent compounds as anequivalent alternative to the FOMC model, with the possibilityto test if n is different from unity.
NAFTA Technical Working Group on Pesticides, Guidance forEvaluating and Calculating Degradation Kinetics inEnvironmental Media (7 p)
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
20/22 jrwb.de
Challenges for the future
Better collaboration
Improve error model
Evaluate related datasets in one step (mixed effect models)
Improve model comparisons (ANOVA, AIC)
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
20/22 jrwb.de
Challenges for the future
Better collaboration
Improve error model
Evaluate related datasets in one step (mixed effect models)
Improve model comparisons (ANOVA, AIC)
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
20/22 jrwb.de
Challenges for the future
Better collaboration
Improve error model
Evaluate related datasets in one step (mixed effect models)
Improve model comparisons (ANOVA, AIC)
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
20/22 jrwb.de
Challenges for the future
Better collaboration
Improve error model
Evaluate related datasets in one step (mixed effect models)
Improve model comparisons (ANOVA, AIC)
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
21/22 jrwb.de
Conclusions
With gmkin + mkin, we now have a completely opensourced software toolset
Biphasic models for metabolites can conveniently be fitted
The use of t-tests for parameter significance is questioned
Plausible confidence intervals for rate constants and singleformation fractions are easily available
(slide corrected after the meeting)
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
21/22 jrwb.de
Conclusions
With gmkin + mkin, we now have a completely opensourced software toolset
Biphasic models for metabolites can conveniently be fitted
The use of t-tests for parameter significance is questioned
Plausible confidence intervals for rate constants and singleformation fractions are easily available
(slide corrected after the meeting)
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
21/22 jrwb.de
Conclusions
With gmkin + mkin, we now have a completely opensourced software toolset
Biphasic models for metabolites can conveniently be fitted
The use of t-tests for parameter significance is questioned
Plausible confidence intervals for rate constants and singleformation fractions are easily available
(slide corrected after the meeting)
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
21/22 jrwb.de
Conclusions
With gmkin + mkin, we now have a completely opensourced software toolset
Biphasic models for metabolites can conveniently be fitted
The use of t-tests for parameter significance is questioned
Plausible confidence intervals for rate constants and singleformation fractions are easily available
(slide corrected after the meeting)
Progress inregulatory
kinetics
PD Dr.Johannes
Ranke
Tasks andmotivation
Confidenceintervals
Metabolitemodels
Conclusions
22/22 jrwb.de
Documentation
kinfit.r-forge.r-project.org/mkin static