QSAR Model Reporting Format

1.QSAR identifier

1.1 QSAR identifier (title)

Molcode QSAR for abiotic degradation in air (OH tropospheric degradation of volatile organic compounds)

1.2 Other related models

1.3 Software coding the model

QSARModel 4.0.4

Molcode Ltd., Turu 2, Tartu, 51014, Estonia
http://www.molcode.com

2.General information

2.1 Date of QMRF

08.02.2010

2.2 QMRF author(s) and contact details

Indrek Tulp
Molcode Ltd.
Turu 2, Tartu, 51014, Estonia
models@molcode.com
http://www.molcode.com

Tarmo Tamm
Molcode Ltd.
Turu 2, Tartu, 51014, Estonia
models@molcode.com
http://www.molcode.com

Gunnar Karelson
Molcode Ltd.
Turu 2, Tartu, 51014, Estonia
models@molcode.com
http://www.molcode.com

Dimitar Dobchev
Molcode Ltd.
Turu 2, Tartu, 51014, Estonia
models@molcode.com
http://www.molcode.com

Dana Martin
Molcode Ltd.
Turu 2, Tartu, 51014, Estonia
models@molcode.com
http://www.molcode.com

Kaido Tämm
Molcode Ltd.
Turu 2, Tartu, 51014, Estonia
models@molcode.com
http://www.molcode.com

Deniss Savchenko
Molcode Ltd.
Turu 2, Tartu, 51014, Estonia
models@molcode.com
http://www.molcode.com

Jaak Jänes
Molcode Ltd.
Turu 2, Tartu, 51014, Estonia
models@molcode.com
http://www.molcode.com

Eneli Härk
Molcode Ltd.
Turu 2, Tartu, 51014, Estonia
models@molcode.com
http://www.molcode.com

Andres Kreegipuu
Molcode Ltd.
Turu 2, Tartu, 51014, Estonia
models@molcode.com
http://www.molcode.com

Mati Karelson
Molcode Ltd.
Turu 2, Tartu, 51014, Estonia
models@molcode.com
http://www.molcode.com

2.3 Date of QMRF update(s)

2.4 QMRF update(s)

2.5 Model developer(s) and contact details

Molcode model development team

Molcode Ltd. Turu 2, Tartu, 51014, Estonia
models@molcode.com
http://www.molcode.com

2.6 Date of model development and/or publication

07.02.2010

2.7 Reference(s) to main scientific papers and/or software package

2.8 Availability of information about the model

Software is proprietary but model training and test sets provided. Algorithm provided.

2.9 Availability of another QMRF for exactly the same model

None to date

3.Defining the endpoint - OECD Principle 1

3.1 Species

Not applicable - environmental fate parameter

3.2 Endpoint

2.Environmental fate parameters. 2.Persistence: Abiotic degradation in air (Phototransformation). 2.2.b.Indirect photolysis (OH-radical reaction, ozone-radical reaction, other)

3.3 Comment on endpoint

Rate constant for OH radical degradation.

The dominant chemical process of chemicals in the gasphase is their reaction with OH radicals, NO₃ radicals, and ozone. The hydroxyl radical is the key reactive species in the troposphere, where it reacts with practically every organic compound.

3.4 Endpoint units

cm³ s^-1 molecule^-1

3.5 Dependent variable

-logK (OH) (original rate constants were transformed into log scale and multiplied by -1 to reduce data range and obtain positive values)

3.6 Experimental protocol

The selected data are for reactions at 25 °C and 1 atm. The gas-phase reaction rate constants of OH radical and organic chemicals have been directly measured.

3.7 Endpoint data quality and variability

Original experimental data were collected from ref 3.

Statistics for -logK(OH):

max value: 15.7

min value: 9.44

standard deviation: 1.03

skewness: 1.16

4.Defining the algorithm - OECD Principle 2

4.1 Type of model

QSAR

4.2 Explicit algorithm

Multilinear regression QSAR
Multilinear regression QSAR derived with BMLR (Best Multiple Linear Regression) method

-logK(OH) = 3.61

+2.15*HASA-1/TMSA (AM1) (all)

-0.698*HOMO energy (AM1)

+1.67*Relative number of aromatic bonds

-12.7*HACA-1/TMSA (Zefirov)

4.3 Descriptors in the model

HASA-1/TMSA (AM1) (all), [unitless]

relative solvent-accessible surface area of H-bonding acceptor atoms (from AM1 calculation)

HOMO energy (AM1), [eV]

energy of highest occupied molecular orbital energy

Relative number of aromatic bonds, [unitless]

Relative number of aromatic bonds

HACA-1/TMSA (Zefirov), [unitless]

sum of solvent-accessible surface area of H-bonding acceptor atoms, selected by threshold charge

4.4 Descriptor selection

Initial pool of ~1000 descriptors for each structure calculated. Stepwise descriptor selection was applied to reduce the pool based on a set of statistical selection rules.

For one-parameter equations: Fisher criterion and R² over threshold, variance and t-test value over threshold, intercorrelation with another descriptor not over threshold)

Two parameter correlations developed from previously reduced pool, the statistical selection applied: intercorrelation coefficient below threshold, significant correlation with endpoint, in terms of correlation coefficient and t-test. Stepwise trial of additional descriptors not significantly correlated to any already in the model. See refs 1-2.

4.5 Algorithm and descriptor generation

1D, 2D, and 3D theoretical calculations. Descriptors derived from mol files. Quantum chemical descriptors from AM1 calculations. Model developed by using multilinear regression using ordinary least squares.

4.6 Software name and version for descriptor generation

QSARModel 4.0.4
QSAR/QSPR package that will compute chemically meaningful descriptors and includes statistical tools for regression modeling
Molcode Ltd, Turu 2, Tartu, 51014, Estonia
http://www.molcode.com

4.7 Chemicals/Descriptors ratio

53 (212 chemicals / 4 descriptors)

5.Defining the applicability domain - OECD Principle 3

5.1 Description of the applicability domain of the model

Applicability domain based on training set:

a) by chemical identity: Diverse set of Volatile Organic Compounds (alphatic and aromatic hydrocarbons, alcohols, amines, halogenated compounds, etc)

b) by descriptor value range: The model is suitable for compounds that have the descriptors in the following minimal-maximal range:

HASA-1/TMSA (AM1) (all): 0 - 0.911

HOMO energy (AM1): -13.3 - -8.10

Relative number of aromatic bonds: 0 - 0.615

HACA-1/TMSA (Zefirov): 0 - 0.0587

5.2 Method used to assess the applicability domain

By chemical identity - compounds must be similar to traing set compounds in terms of functionality.

By descriptor value range: range of descriptor values similar to training set with ±30% confidence. Descriptor values must fall between maximal and minimal descriptor values of training set ±30%.

5.3 Software name and version for applicability domain assessment

QSARModel 4.0.4
QSAR/QSPR package that will compute chemically meaningful descriptors and includes statistical tools for regression modeling
Molcode Ltd, Turu 2, Tartu, 51014, Estonia
http://www.molcode.com

5.4 Limits of applicability

See 5.1

6.Internal validation - OECD Principle 4

6.1 Availability of the training set

Yes

6.2 Available information for the training set

Chemname:Yes
SMILES:No
CAS RN:Yes
InChI:No
MOL file:Yes
Formula:Yes

6.3 Data for each descriptor variable for the training set

All

6.4 Data for the dependent variable for the training set

All

6.5 Other information about the training set

212 data points: 0 negative values; 212 positive values

original source dataset split into testing and training. From the original source data of 423 values sorted by endpoint value, each 2nd was subjected to the test set.

6.6 Pre-processing of data before modelling

No more than specified in 3.5

6.7 Statistics for goodness-of-fit

R² = 0.832 (Correlation coefficient)

s₂ = 0.427 (Standard error of the estimate)

F = 256.8 (Fisher function)

6.8 Robustness - Statistics obtained by leave-one-out cross-validation

R²_CV = 0.821

6.9 Robustness - Statistics obtained by leave-many-out cross-validation

R²_CVMO = 0.819 (80% : 20% , training : testing)

6.10 Robustness - Statistics obtained by Y-scrambling

6.11 Robustness - Statistics obtained by bootstrap

6.12 Robustness - Statistics obtained by other methods

ABC analysis (2:1 training : prediction) on sorted (in increased order of endpoint value) data divided into 3 subsets (A;B;C). Training set formed with 2/3 of the compounds (set A+B, A+C, B+C) and validation set consisted of 1/3 of the compounds (C, B, A).

average R² (fitting) = 0.833

average R² (prediction) = 0.824

7.External validation - OECD Principle 4

7.1 Availability of the external validation set

Yes

7.2 Available information for the external validation set

Chemname:Yes
SMILES:No
CAS RN:Yes
InChI:No
MOL file:Yes
Formula:Yes

7.3 Data for each descriptor variable for the external validation set

All

7.4 Data for the dependent variable for the external validation set

All

7.5 Other information about the external validation set

211 data points: 0 negative values; 211 positive values

7.6 Experimental design of test set

Original source dataset split into testing and training. From the original source data, sorted by endpoint value, each 2nd was subjected to the test set.

7.7 Predictivity - Statistics obtained by external validation

R² = 0.773 (Coefficient of determination)

7.8 Predictivity - Assessment of the external validation set

All are in range of applicability domain:

HASA-1/TMSA (AM1) (all): 0 - 0.942

HOMO energy (AM1): -13.1 - -8.27

Relative number of aromatic bonds: 0 - 0.579

HACA-1/TMSA (Zefirov): 0 - 0.0561

7.9 Comments on the external validation of the model

The validation coefficient of determination (R²) is significant and close to the coefficients derived by internal validation (R²_CV and R²_CVMO).

8.Providing a mechanistic interpretation - OECD Principle 5

8.1 Mechanistic basis of the model

The descriptors "HASA-1/TMSA (AM1) (all)" and "HACA-1/TMSA (Zefirov)" are simultaneously taking into account the capability of hydrogen acceptor bonding and the size of the compound. Although the descriptors seem to be similar, they are counting different features of hydrogen acceptor abilities. "HASA-1/TMSA (AM1) (all)" counts all possible hydrogen acceptor atoms solvent accessible surface area while "HACA-1/TMSA (Zefirov)" counts only charged areas. "HOMO energy (AM1)" is an indicator of the nucleophilicity of the molecule - reactive molecules have relatively higher HOMO energy.. "Relative number of aromatic bonds" represents a count of aromaticity which differentiates these compounds from aliphatic ones. All presented descriptors represent important molecular properties related to H-abstraction.

8.2 A priori or a posteriori mechanistic interpretation

A posteriori mechanistic interpretation, consistent with published scientific interpretations of experiments (in ref. 4. HOMO energy and aromatic carbon were found to be important)

8.3 Other information about the mechanistic interpretation

9.Miscellaneous information

9.1 Comments

Similar methodology to current approach was used in refs 1-2

9.2 Bibliography

Karelson M, Dobchev D, Tamm T, Tulp I, Jänes J, Tämm K, Lomaka A, Savchenko D & Karelson G (2008). Correlation of blood-brain penetration and human serum albumin binding with theoretical descriptors. ARKIVOC 16, 38-60.

Karelson M, Karelson G, Tamm T, Tulp I, Jänes J, Tämm K, Lomaka A, Savchenko D & Dobchev D (2009). QSAR study of pharmacological permeabilities. ARKIVOC 2, 218–238.

Atkinson R (1989). Kinetics and mechanisms of the gas-phase reactions of the hydroxyl radical with organic compounds. Journal of Physical Reference Data, Monograph 1.

Gramatica P, Pilutti P & Papa E (2004). Validated QSAR Prediction of OH Tropospheric Degradation of VOCs: Splitting into Training?Test Sets and Consensus Modeling. Journal of Chemical Informatics and Computer Science 44, 1794–1802.

9.3 Supporting information

Training data set	Photolysis_#1_212_training
Validation data set	Photolysis_# 1_211_testset
Other documents

10.Summary (ECB Inventory)

10.1 QMRF number

10.2 Publication date

10.3 Keywords

Molcode, abiotic degradation in air, OH tropospheric degradation, volatile organic compounds

10.4 Comments