Registration Dossier

Administrative data

Endpoint:
melting point/freezing point
Type of information:
(Q)SAR
Adequacy of study:
weight of evidence
Reliability:
2 (reliable with restrictions)
Rationale for reliability incl. deficiencies:
results derived from a (Q)SAR model, with limited documentation / justification, but validity of model and reliability of prediction considered adequate based on a generally acknowledged source
Justification for type of information:
1. Software tool(s) used including version: EPI SuiteTM v4.11

2. Model(s) used: MPBPWIN v1.43

3. SMILES OR OTHER IDENTIFIERS USED AS INPUT FOR THE MODEL :: C(=O)(CCCCCCCCCCC(CCCCCC)O)OCC(CCCC)CC

4. SCIENTIFIC VALIDITY OF THE (Q)SAR MODEL
*Estimation Methodology*

MPBPWIN estimates melting point by two different methods.  The first is an adaptation of the Joback group contribution method for melting point (Joback, 1982; Reid et al; 1987) and the second is a simple Gold and Ogle method suggested by Lyman (1985).
The original Joback methodology used a data set of 388 compounds to derive 41 chemical structure group descriptors via multiple linear regression (Joback, 1982).  The Joback adaptation in MPBPWIN is an extension of the original method to include the same groups as in the adapted Stein and Brown boiling point method (see Boiling Point).  In addition, MPBPWIN also uses melting point correction factors for specific structures. Appendix F contains a complete list the group descriptors and coefficient values.
The second estimation method (Gold and Ogle, 1969), simply relates melting point (Tm) to boiling point (Tb) as follows (both values in K):
Tm  =  0.5839 Tb
MPBPWIN averages the adapted Joback and the Gold and Ogle estimates and reports the average estimate as well as both individual estimates.
MPBPWIN then goes one step further.  It reports a "suggested" melting point (MP) that is based upon the two individual estimates and several criteria.  First, MPBPWIN looks at the difference between the two estimates.  If the difference is small (< 30 K), the suggested MP is simply the average.  When this criteria fails (which occurs quite often), MPBPWIN examines the structure type and the magnitude of the difference.  It then decides which estimate is more likely to be accurate and "weights" the suggested MP accordingly.  For example, when MPBPWIN detects an amino-acid structure, it uses a 75% weighting factor for the higher estimate and 25% for the lower estimate to derive the suggested MP.  Weighting factors in MPBPWIN were approximated through observation of estimated versus experimental MP.

The adapted Joback method can significantly over-estimate MP for some structures.  A similar error occurs in the Stein and Brown (1994) boiling point method (when BP > 500 K) before a  quadratic or linear equation corrects the error.  This type of correction was not developed for MPBPWIN.  Instead, MPBPWIN applies a "cut-off" MP at approximately 350 deg C; that is, any MP estimate above 350 deg C is reduced to 350 deg C.  When MPBPWIN detects a large difference between a very high adapted Joback estimate and a much lower Gold and Ogle estimate, it usually weights the suggested MP strongly to the Gold and Ogle estimate (again, it depends on structure).  When used alone, the adapted Joback MP method can be very inaccurate for some structures (usually by estimating too high).  The simplistic Gold and Ogle method is also inaccurate for various structures.  However, when combined in the MPBPWIN format, estimation accuracy  improves significantly for very large, diverse datasets.

*Estimation Accuracy*

Although the suggested MPBPWIN estimates may be adequate for screening purposes, the overall accuracy is not outstanding.  In fact, most current methods for estimating MP (for large diverse datasets) have generally poor accuracy and can yield many unreliable estimates (Lyman, 1985; Reid et al, 1987).  The failure derives, in part, from ignoring the effects of symmetry in the molecule (Lyman, 1985).

The original Joback method is based on a relatively small training set of 388 compounds with the following accuracy statistics (Joback, 1982; Reid et al, 1987).
number = 388
standard deviation = 25 K
average deviation  = 23 K
absolute error = 11%

The adapted method in MPBPWIN does not affect these statistics since the original method had already defined the necessary group fragments.  The method error is comparable to other MP estimation methods for these types of simple structures (haloalkanes, plain hydrocarbons, acids, esters, alcohols, ethers, ketones, simple aromatics, etc).  However, application of the original Joback method (or other estimation methods) to a large dataset of more diverse compounds (such as pesticides, drugs, multi-functional aromatics, etc.) produces much higher error.

For the current EPI Suite, the accuracy of the "suggested" MPBPWIN melting point estimate was tested on a large dataset of 10,051 compounds containing a diverse mix of simple, moderate and very complex structural compounds (includes many pesticides and pharmaceutical compounds).  The dataset was taken from the PHYSPROP Database used by the EPI Suite.  Compounds having "decompose" designations with MP values were excluded.  The complete dataset with experimental values and estimates is available at:

http://esc.syrres.com/interkow/EpiSuiteData.htm

5. APPLICABILITY DOMAIN
*Estimation Domain*

Currently there is no universally accepted definition of model domain.  However, users may wish to consider the possibility that property estimates are less accurate for compounds outside the Molecular Weight range of the training set compounds, and/or that have more instances of a given fragment than the maximum for all training set compounds.  It is also possible that a compound may have a functional group(s) or other structural features not represented in the training set, and for which no fragment coefficient was developed.  These points should be taken into consideration when interpreting model results.

The complete training sets for MPBPWIN's estimation methodology are not available.  Therefore, describing a precise estimation domain for this methodology is not possible.

The original Joback methodology was developed with only 388 compounds.  The number of compounds used to derive each of the original Joback descriptors is shown in Appendix F, but the maximum number of each descriptor that occurs in each compound is not available.  Similar data for the Gold and Ogle method is not available.

The current applicability of the MPBPWIN methodology is best described by its accuracy in predicting melting point.  The complete dataset with experimental values and estimates is available via Internet download: http://esc.syrres.com/interkow/EpiSuiteData.htm

This test set probably contains more than 15 times the number of compounds used to train the methodology.  As can be seen in the Accuracy section above, significant errors are possible.


6. ADEQUACY OF THE RESULT
The result is considered adeguate for C&L procedure and for risk assessment.

Data source

Reference
Reference Type:
publication
Title:
EPISUITE v.4.11 - MPBPWIN™
Year:
2018
Bibliographic source:
MPBPWIN™ - US EPA. [2018]. Estimation Programs Interface Suite™ for Microsoft® Windows, v 4.11. United States Environmental Protection Agency, Washington, DC, USA.

Materials and methods

Test guidelineopen allclose all
Qualifier:
according to
Guideline:
other: REACH Guidance on QSARs R.6
Version / remarks:
2008
Deviations:
not applicable
Qualifier:
according to
Guideline:
other: Practical guide How to use and report (Q)SARs (ECHA)
Version / remarks:
Version 3.1 – July 2016
Deviations:
not applicable
Principles of method if other than guideline:
- Software tool(s) used including version:
EPI SuiteTM v4.11
- Model(s) used:
MPBPWIN v1.43
- Model description: see field 'Justification for non-standard information'
- Justification of QSAR prediction: see field 'Justification for type of information'
GLP compliance:
no
Remarks:
not relevant
Other quality assurance:
other: not relevant
Remarks:
(Q)SAR prediction
Type of method:
other: (Q)SAR prediction

Test material

Reference
Name:
Unnamed
Type:
Constituent
Test material form:
other: not applicable for in silico study
Details on test material:
- State of aggregation:
not specified
- Particle size distribution:
not specified
- Mass median aerodynamic diameter (MMAD):
not specified
- Geometric standard deviation (GSD):
not specified
- Shape of particles:
not specified
- Surface area of particles:
not specified
- Crystal structure:
not specified
- Coating:
not specified
- Surface properties:
not specified
- Density:
not specified
- Moisture content:
not specified
- Residual solvent:
not specified
- Activation:
not specified
- Stabilisation:
not specified
- Other: not specified
Specific details on test material used for the study:
SMILE NOTATION: C(=O)(CCCCCCCCCCC(CCCCCC)O)OCC(CCCC)CC
CHEM : 2-Ethylhexyl 12-hydroxystearate
MOL FOR: C26-H52-O3
MOL WT : 412.70

Results and discussion

Melting / freezing pointopen allclose all
Key result
Melting / freezing pt.:
167.66 °C
Remarks on result:
other: Mean Value
Remarks:
(Q)SAR prediction value
Melting / freezing pt.:
182.77 °C
Remarks on result:
other: Adapted Joback Method
Remarks:
(Q)SAR predicted value

Any other information on results incl. tables

Experimental Database Structure Match: no data

Applicant's summary and conclusion

Conclusions:
A prediction was performed by EPISUITE v.4.11 - MPBPWIN™ for 2-ethylhexyl 12-hydroxyoctadecanoate for the determination of the melting point of the substance and it resulted equal to 167.66°C (mean value).
Executive summary:

A prediction was performed by EPISUITE v.4.11 - MPBPWIN™ for 2-ethylhexyl 12-hydroxyoctadecanoate for the determination of the melting point of the substance and it resulted equal to 167.66°C (mean value).

Adequacy of the QSAR:

1) QSAR model is scientifically valid.

2) The substance falls within the applicability domain of the QSAR model.

3) The prediction is fit for regulatory purpose. The prediction is adequate for the Classification and Labelling or risk assessment of the substance as indicated in REACH Regulation (EC) 1907/2006: Annex XI Section 1.3

4) The information related to the model applied, the results obtained are well documented.