The CORAL is freeware to build up quantitative structure-property activity relationships (QSPR /QSAR).

The CORAL can be used in Windows platforms.

The CORAL gives possibility to build up models for endpoints related to nanomaterials.

The model gives possibility to estimate (predict) numerical value of endpoint for a substance that has not been examined in experiment.

One can calculate by the least squares method a model E = C0 + C1*D.

These QSPR/QSAR models are one-variable correlations between an endpoint and descriptors calculated with SMILES (simplified molecular input line entry system).

CORALis system of software and databases. Input for CORAL is data set in form of strings.

Each string contains four components:

[TypeSet. ID. SMILES.  Endpoint ]

TypeSet :+/-is an indicator of sub-training/calibration sets ; #  is an indicator of test set.

ID: is some indicator of a substance, e.g. CAS number.

SMILES: is, e.g. CCC(C), N#CC, c1cccc1, etc.

quasi-SMILES: quasi-SMILES contains fragment which is the traditional SMILES. Quasi- SMILES can be defined with sequences of symbols in square brackets, e.g. [aAl2O3][b39,7][c267][d36,3][e64,7][f-17,345][g-1,51][h-9,81][i5,67][jMTT][kHCMEC][LHuman][mBlood][nNormal][o24][p0,001]; or O=S(=O)(c1ccccn1=O)Cc2ccccc2.[Dark][Low Conc] etc.

Endpoint: is some numerical value, e.g. 0.15; 2.78, -134.7, etc.

Using sub-training set and calibration set, the CORAL is building up a model for an endpoint:

Endpoint = C₀ + C₁*Descriptor(SMILES)

Endpoint = C₀ + C₁*Descriptor(quasi-SMILES)

Formally, the model can be used for arbitrary external substances, but preferably if the external substance falls into the applicability domain.