Bioconcentration factor (logBCF) is an important ecological indicator of a substance. Experimental definition of the logBCF for all substances is impossible. The CORAL software was tested as a potential tool for modeling of bioconcentration factor. Using this software the data was analyzed in three runs of the program. In three random splits the data was divided into the sub-training (≈45%), the calibration (≈34%), and the test sets (≈22%). The obtained models are characterized by the following statistical quality: Split 1: n=216, r 2=0.839, s=0.54, F=1115 (sub-training); n=188, r 2=0.839, s=0.54 (calibration); n=118, r 2=0.878, s=0.50 (test); Split 2: n=241, r 2=0.830, s=0.56, F=1169 (sub-training); n=167, r 2=0.871, s=0.51 (calibration); n=114, r 2=0.881, s=0.46 (test); Split 3: n=244, r 2=0.835, s=0.57, F=1228 (sub-training); n=171, r 2=0.847, s=0.53 (calibration); n=107, r 2=0.860, s=0.49 (test). Structural features which can be extracted from simplified molecular input line entry system (SMILES) and which are statistically significant promoters of increase or promoters of decrease for bioconcentration factor are identified and discussed.
- Bioconcentration factor
- CORAL software
- OECD principles
ASJC Scopus subject areas
- Analytical Chemistry
- Computer Science Applications
- Process Chemistry and Technology