CPANNatNIC software for counter-propagation neural network to assist in read-across

Drgan et al. J Cheminform CPANNatNIC software for counter-propagation neural network to assist in read-across Viktor Drgan 0 2 Špela Župerl 0 2 Marjan Vračko 0 2 Claudia Ileana Cappelli 1 Marjana Novič 0 2 0 Department of Cheminformatics, National Institute of Chemistry , Hajdrihova 19, 1001 Ljubljana , Slovenia 1 Laboratory of Environmental Chemistry and Toxicology, IRCCS-Istituto di Ricerche Farmacologiche Mario Negri , Via La Masa 19, Milan , Italy 2 Department of Cheminformatics, National Institute of Chemistry , Hajdri- hova 19, 1001 Ljubljana , Slovenia Background: CPANNatNIC is software for development of counter-propagation artificial neural network models. Besides the interface for training of a new neural network it also provides an interface for visualisation of the results which was developed to aid in interpretation of the results and to use the program as a tool for read-across. Results: The work presents the details of the program's interface. Parts of the interface are presented and how they can be used. The examples provided show how the user can build a new model and view the results of predictions using the interface. Examples are given to show how the software may be used in read-across. Conclusions: CPANNatNIC provides a simple user interface for model development and visualisation. The interface implements options which may simplify read-across procedure. Statistical results show better prediction accuracy of read-across predictions than model predictions where similar compounds could be identified, which indicates the importance of using read-across and usefulness of the program. Counter-propagation neural network; Read-across; Software - Background In the past several years, there is an increasing interest in using in silico tools for risk assessment of chemicals. The reasons for higher interest can be found in Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH) legislation in European Union which requires registration of a large number of chemicals in use. The legislation allows using read-across for toxicity assessment under certain conditions written in the regulation. Definition of read-across and its correct use are still rather unclear. Patlewicz et  al. [1] gathered several definitions of read-across from different sources [e.g. United States Environmental Protection Agency (US EPA), European Chemical Agency (ECHA), The Organisation for Economic Co-operation and Development (OECD)]. Concisely, we may understand the definitions of read-across as an approach to predict a property of a chemical based on the same property of one or more similar chemicals. Different tools already exist which can be used for read-across, for example OECD QSAR Toolbox [2], ToxRead [3], TEST [4] and VEGA [5]. In this paper we present a new tool which can be used for development of counter-propagation artificial neural network (CPANN) models. The models can be later used either for direct prediction of the endpoint under consideration for new, i.e. untested compounds, or for read-across approach. The software provides a graphical user interface which was designed to facilitate readacross based on analogue or category approach using CPANN models. CPANNs are particularly suitable for these approaches because of their ability to group compounds according to their structural similarity. Although the software was initially built to facilitate read-across for toxicity assessment of substances, its usage is not limited to toxicity-related endpoints since the user describes compounds in the input data file(s) which may include numerical values of any property. © The Author(s) 2017. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/ publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Basis for read‑across As mentioned above, the software uses CPANN models. The results of the predictions can be viewed in a simple graphical user interface with compounds placed on the map, called a “top-map”, according to their similarity which can be used as the basis for read-across predictions. The learning principles of Kohonen and CPANNs are well established and can be found in detail elsewhere [6–8]. Some definitions are given below so that the user can better understand the results produced by the software. Schematic representation of a CPANN is shown in Fig. 1. It is composed of Kohonen layer and output (Grossberg) layer. It can be visualized as a 3D matrix of values called weights (W). One column (vector) of weights is call (...truncated)