An open source multistep model to predict mutagenicity from statistical analysis and relevant structural alerts

BMC Chemistry

Table 2 Confusion matrix of mutagenicity integrated model on the training set (3367 chemical compounds).

Training set (3367 chemicals)	Mutagenic predictions	Non-mutagenic predictions	Suspicious predictions	Unpredicted compounds
Mutagens	1798	69	15	1
Non-mutagens	169	1239	76	0

The low number of true positives in the suspicious set, if compared with the test set confusion matrix (cf. Table 1), is due to the very small number of real mutagens in the "non-mutagenic" predictions on the training set. The unpredicted structure was not processed by the CDK library.

ISSN: 2661-801X