Using Biowin™, Bayes, and batteries to predict ready biodegradability

Abstract

Whether or not a given chemical substance is readily biodegradable is an important piece of information in risk screening for both new and existing chemicals. Despite the relatively low cost of Organization for Economic Cooperation and Development tests, data are often unavailable and biodegradability must be estimated. In this paper, we focus on the predictive value of selected Biowin models and model batteries using Bayesian analysis. Posterior probabilities, calculated based on performance with the model training sets using Bayes' theorem, were closely matched by actual performance with an expanded set of 374 premanufacture notice (PMN) substances. Further analysis suggested that a simple battery consisting of Biowin3 (survey ultimate biodegradation model) and Biowin5 (Ministry of International Trade and Industry [MITI] linear model) would have enhanced predictive power in comparison to individual models. Application of the battery to PMN substances showed that performance matched expectation. This approach significantly reduced both false positives for ready biodegradability and the overall misclassification rate. Similar results were obtained for a set of 63 pharmaceuticals using a battery consisting of Biowin3 and Biowin6 (MITI nonlinear model). Biodegradation data for PMNs tested in multiple ready tests or both inherent and ready biodegradation tests yielded additional insights that may be useful in risk screening.

Biodegradation and removal of pharmaceuticals and personal care products in treatment systems: a review

“…The regression models have been shown to possess the highest utility and are currently being adapted by the U.S. Environmental Protection Agency as EPA BIOWIN predictive models, but AI approaches have recently gained attention for their potential to greatly improve prediction accuracy (Klopman and Tu 1997;Rorije et al 1999;Baker et al 2004). The majority of current regression models rely mainly on structure activity relationships, in which statistical models (mostly regressions or Bayesian statistics) are applied based on expert knowledge regarding the biodegradability of organic compounds according to their structures (Boethling et al 2004). Advances in the understanding of structure-biodegradability relationships have provided valuable information regarding the biodegradability of PPCPs.…”

“…They found that certain compounds, including esters, nitriles, and aromatic alcohols, have functional groups that usually increase a compound's biodegradability, whereas aromatic amines, iodide, nitro, and azo groups tend to render a compound more recalcitrant (Tunkel et al 2000). Boethling et al (2004) have demonstrated the utility of the computational model by predicting the readily biodegradable nature of 63 pharmaceuticals with reasonable accuracy (83% and 87%) using BIOWIN 5 and 6 models. However, though current prediction models have proven useful, inconsistencies between different models and inaccuracies have been observed.…”

A methodology for ranking and hazard identification of xenobiotic organic compounds in urban stormwater

“…If these data are not available, but degradation half-lives have been reported, the latter are used according to the criteria listed in Table 1. If no half-lives are available, estimation of the biodegradability is made using the BioWin software included in the EpiSuite program (US-EPA, 2005) following the recommendations made by Boethling et al (2004) as outlined in Table 1.…”

Review of Estimation Models for Biodegradation

“…In BIOWIN version 4.02, a battery use of models was added to give a qualitative (yes/no) prediction for ready biodegradability [48]. The criteria are as follows: if the Biowin3 (ultimate survey model) result is weeks or faster (e.g.…”