QSAR Model Reporting Format

1.QSAR identifier

1.1 QSAR identifier (title)

Nonlinear QSAR: artificial neural network for dermal irritation

1.2 Other related models

http://reachqsar.com/

1.3 Software coding the model

QSARModel 3.3.8

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

Statistica 7

StatSoft Ltd.
http://www.statsoft.com/

2.General information

2.1 Date of QMRF

10.10.2010

2.2 QMRF author(s) and contact details

Dimitar Dobchev
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

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

Indrek Tulp
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

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
www.molcode.com

2.6 Date of model development and/or publication

12.04.2010

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

Katritzky A R, Dobchev DA, Fara DC, Hur E, Tämm K, Kuruncz L, Karelson M, Varnek A & Solov'ev VP (2006). Skin Permeation Rate as a Function of Chemical Structure. Journal of Medicinal Chemistry 49, 3305-3314.

Karelson M, Dobchev DA, Kulshyn OV & Katritzky A (2006). Neural Networks Convergence Using Physicochemical Data. Journal of Chemical Information and Modeling 46, 1891- 1897.

Statistica 7

2.8 Availability of information about the model

Selection, training and test sets are available. Model algorithm is available (snn file).

2.9 Availability of another QMRF for exactly the same model

None to date.

3.Defining the endpoint - OECD Principle 1

3.1 Species

Rabbit

3.2 Endpoint

4.Human health effects. . 4.4.Skin irritation /corrosion

3.3 Comment on endpoint

Dermal irritation is the production of reversible inflammatory changes in irritation skin following the application of a substance. The skin irritation potential is described by the Primary Irritation Index (PII), calculated from erythema and oedema grades based on experimental rabbits. The maximum PII is 8 and the minimum is 0. The grading scale for irritant effects on rabbit skin were originally proposed by Draize and adopted by the OECD (Test Guideline 404) and the US and EU regulatory agencies [ref 1, sect 9.2].

The PII can be calculated as:

PII = [SUM(Erythema 24/48/72 h)+SUM(Oedema24/48/72 h)] (3 x no. animals) where Erythema is redness of skin produced by vascular congestion or increased perfusion And Oedema is the presence of abnormally large amounts of fluid in the intercellular tissue spaces of the epidermis, dermis or subcutaneous tissues.

3.4 Endpoint units

3.5 Dependent variable

Primary Irritation Index (PII)

3.6 Experimental protocol

All 286 data used in this report were obtained from in vivo rabbit skin irritation test that were used to assess the potential of materials to cause skin irritancy or corrosion in man, and to meet regulatory requirements which require classification and appropriate labeling of a material if it is believed to be potential irritant or corrosive. All chemicals were tested applying a volume or weight of 0.5ml or 0.5g undiluted, except where an alternative weight or concentration was needed. Exposure time for each test was 4 hours [ref 1, sect 9.2].

3.7 Endpoint data quality and variability

The 286 chemicals selected were readily available at high and consistent purity and are expected to be stable on storage. They have been tested undiluted in in vivo studies, excepting those chemical where high concentrations of the substance could be expected to cause sever effects. The invivo data were generated in 1981 in studies carried out according to OECD Test Guideline 404 and following the principles of Good Laboratory Practice. The data presented were obtained from tests normally using at least three rabbits involving application of 0.5 ml (or 0.5g) to the flank under semi-occlusive patches and in which observations were made at least 24, 48 and 72 hours [ref 1, sect 9.2].

4.Defining the algorithm - OECD Principle 2

4.1 Type of model

Neural network

4.2 Explicit algorithm

Neural network
Nonlinear QSAR: backpropagation Neural Network (Multilayer Perceptron) regression

The algorithm is based on regression neural network predictor with structure 9-8-6-1.

4.3 Descriptors in the model

count of H-acceptor sites (AM1) ,

HBCA H-bonding charged surface area (AM1) ,

FHACA Fractional HACA (HACA/TMSA) (AM1) ,

Average atom weight ,

min(#HA, #HD) (AM1) ,

HACA-2 (AM1) ,

Number of O atoms ,

Difference (Pos - Neg) in Charged Surface Areas (Zefirov) ,

Negatively Charged Part of Charged Surface Area (AM1) ,

4.4 Descriptor selection

Initial pool of ~1000 descriptors. Stepwise descriptor selection based on a set of statistical selection rules as F statistic and p. The first highest F (low p) descriptors (9) were selected from the whole set of descriptors. These 9 descriptors were used as inputs to the network. 29 networks with different structures were tested in order to find the best ANN with lowest RMS (root-mean-squared error) and highest correct predictions (for training, selection and test sets). Then 245 epochs were used to train the final network with architecture depicted in 4.2. Optimization of the weights was performed with Levenberg-Marquardt algorithm encoded in the backpropagation scheme using linear and hyperbolic activation functions.

4.5 Algorithm and descriptor generation

All descriptors were generated using QSARModel on structure optimized by AM1 semiempirical quantum mechanical model. The final structure were optimized by mopac6 implemented in QSARModel. Keywords used for opyimizations were: AM1 EF GNORM=0.05 BONDS PI POLAR ENPART NOINTER PRECISE. The final descriptors were selected as denoted in 4.4 as well as descriptors with small variances less than 10 e-⁵ were discarded from the total pool.

4.6 Software name and version for descriptor generation

QSARModel 3.3.8


http://www.molcode.com

4.7 Chemicals/Descriptors ratio

16.2 ( 146 chemicals / 9 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:

By descriptor value range (between min and max values): The model is suitable for compounds that have the descriptors in the following range augmented with the confidence in 5.2:

Desc ID

See 4.3: 1 2 3 4 5 6 7 8 9

Min: 0.000 0.000 0.000 4.588 0.000 0.000 0.000 -243.84 1.705

Max: 4.000 92.304 0.209 27.639 4.000 6.156 6.000 952.967 140.451

5.2 Method used to assess the applicability domain

Presence of functional groups in structures (ethers, esters, amides, halides, aromatic, aliphatic functional groups etc)

Range of descriptor values in training set with ±30% confidence

Descriptor values must fall between maximal and minimal descriptor values (see5.1) of training set ±30%.

5.3 Software name and version for applicability domain assessment

QSARModel 3.3.8


http://www.molcode.com

5.4 Limits of applicability

See 5.1, 5.2

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:No

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

Data points: 146

6.6 Pre-processing of data before modelling

Standardization and normalization of the inputs by taking into account the mean and standard deviation

6.7 Statistics for goodness-of-fit

Training PII; Selection PII; Test PII

Data Mean: 2.348; 3.129; 2.417

Data SD: 2.040; 2.512; 1.610

Error Mean: -0.019; -0.009; -0.222

Error SD: 1.185; 2.781; 1.390

Abs E. Mean: 0.845; 1.903; 1.112

S.D. Ratio: 0.581; 1.107; 0.864

Correlation: 0.814; 0.628; 0.590

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

See 6.7

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

6.10 Robustness - Statistics obtained by Y-scrambling

6.11 Robustness - Statistics obtained by bootstrap

6.12 Robustness - Statistics obtained by other methods

RMS(Training)=0.14814; RMS (Selection)=0.347624; RMS(Test)=0.176003

In this ANN 2 sets of randomly chosen (20) data to test the network – selection set and test set, See also 6.7

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:No

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

The method used two validation sets: selection (20) and test (20)

7.6 Experimental design of test set

Randomly selected 20 selection and 20 test data points

7.7 Predictivity - Statistics obtained by external validation

See 6.7 and 6.12

7.8 Predictivity - Assessment of the external validation set

The descriptors for the test set are in the limit of applicability, see 6.7 and 6.12. We have limited ourselves to select two auxiliary sets to train the network and to test it externally on the test set. Thus more than 1.5 of the datapoints were selected for these two sets divided by 2. One of the main purposes of the ANN model also to be applicable for diverse compounds for future predictions, thus we wtried to keep the training set as large as possible and to select the validation and test sets with significant data points.

7.9 Comments on the external validation of the model

Overall predictions for the selection set (used to stop the ANN training and not to overfit it) and the test set (used to test the external prediction of the net after training) are significant according to the RMS error and the standard deviation ratio (S.D.Ratio); see 6.7 and 6.12.

8.Providing a mechanistic interpretation - OECD Principle 5

8.1 Mechanistic basis of the model

The complex nature of the ANN model does not allow direct interpretation of the descriptors in relation to the modelled property. However, it can be noted that descriptors related to the hydrogen bonding ability and the charged surface areas of the molecules are mainly present. The reactivity of the compounds with the epidermis depends also on charged surface areas of the compounds (which are the most reactive sides). Several authors have confirmed the reactivity related with the charged surfaces and also the LUMO and HOMO descriptors [ref 2,3; sect 9.2]. It can be roughly estimated that the PII increases with increasing (slight negative correlation between the descriptors) count of H-acceptor sites (AM1), HBCA H-bonding charged surface area (AM1), and FHACA Fractional HACA (HACA/TMSA) (AM1).

8.2 A priori or a posteriori mechanistic interpretation

8.3 Other information about the mechanistic interpretation

9.Miscellaneous information

9.1 Comments

Supporting information for: training set(s), selection set(s), test set(s).

The 9-8-6-1.snn file includes the ANN model; the user must have Statistica 7 or higher with ANN modules to make predictions.

9.2 Bibliography

ECETOC Technical Report No 66. Skin Irritation and Corrosion: Reference Chemicals Data Bank. March 1995

Kodithala K, Hopfinger AJ, Thompson ED & Robinson MK (2002). Prediction of skin irritation from organic chemicals using membrane-interaction QSAR analysis. Toxicological Sciences 66, 336–346.

Hayashi M, Nakamura Y, Higashi K, Kato H, Kishida F & Kaneko H (1999). A quantitative structure-Activity relationship study of the skin irritation potential of phenols. Toxicology in Vitro 13, 915-922.

9.3 Supporting information

Training data set	Dermal_Irritation_PII_training_ 146.sdf
Validation data set	Dermal_Irritation_PII_selection_20.sdf Dermal_Irritation_PII_test_20.sdf
Other documents	9-8-6-1.snn

10.Summary (JRC Inventory)

10.1 QMRF number

Q17-22-1-332

10.2 Publication date

2011/12/19

10.3 Keywords

skin irritation, PII, Draize, Molcode

10.4 Comments

To be entered by JRC