Paola Gramatica on Quantitative Structure-Activity Relationship Models
Fast Moving Front Commentary, January 2011
![]() |
Article: Statistical external validation and consensus modeling: A QSPR case study for K-oc prediction
Authors: Gramatica, P;Giani, E;Papa,
E |
Paola Gramatica talks with ScienceWatch.com and answers a few questions about this month's Fast Moving Fronts paper in the field of Computer Science.
Why do you think your paper is highly
cited?
This paper is an example of the possibility to predict data, not experimentally available, starting from a limited amount of good experimental data for soil sorption partition coefficient (Koc), by applying well-validated Quantitative Structure-Activity Relationship (QSAR) models. In Silico QSAR models exploit the experimental information available for a limited amount of chemicals. They are based on structural molecular descriptors and statistical methods and can predict data for new or not-tested chemicals.
Does it describe a new discovery, methodology, or
synthesis of knowledge?
The novelty of this work consists in the strong validation performed during the development of the QSAR models: only 93 out of the experimental LogKoc data available for 643 chemicals were used as input for the development of QSAR models. The good accordance between the experimental data for the other 550 chemicals and the data predicted by our models is a proof of the reliability of the QSAR predictions. Additionally, predictions by consensus based on different QSAR models of similar high quality are proposed.
Would you summarize the significance of your paper
in layman's terms?
"The main implications of my research, as well as of all QSAR works, is the possibility to screen big data sets, to prioritize the more dangerous and to focus the experimental tests only on them."
The significance of any valid QSAR model is the possibility to apply it for obtaining not experimentally available data: the models which were developed, validated, and proposed in this paper can be applied with high confidence for the prediction of soil sorption partition coefficient of organic compounds (LogKoc). This parameter is highly important to predict soil sorption or leaching to water of pesticides, but also of any other organic chemical in the environment. Furthermore, the evaluation of the applicability domain of the proposed QSAR model makes it possible to know for which chemicals it can be reliably applied.
How did you become involved in this research, and
how would you describe the particular challenges, setbacks, and
successes that you've encountered along the way?
I've been a QSAR model developer for the last 15 years and I'm mainly interested in studies on environmental pollution by organic compounds. I have published about 100 papers on QSAR models on several topics in this field, all highly cited in international literature (h index: 28, more than 2,000 citations; this paper is 16 in the citations list of my papers). As a recognized researcher in QSAR modeling of environmental chemicals I'm often invited to present my works in international meetings and to write book chapters.
Where do you see your research leading in the
future?
I hope that all my future works will have the success of my previous ones: the validation of QSAR models is my principal focus. Only validated models can produce reliable predicted data.
Do you foresee any social or political
implications for your research?
The main implication of my research, as well as of all QSAR works, is the
possibility to screen large data sets, to prioritize the most dangerous
chemicals and to focus the experimental tests only on them. This approach
has, as an important consequence, the possibility to reduce animal tests
and the general cost of experiments. QSAR models are also requested in the
new European legislation of chemicals, REACH (Registration, Evaluation,
Authorisation and restriction of CHemicals).
Prof. Paola Gramatica
Full Professor of Environmental Chemistry
- past Associate Professor of Organic Chemistry
School of Sciences- University of Insubria
Department of Structural and Functional Biology (DBSF)
QSAR Research Unit in Environmental Chemistry and Ecotoxicology
Varese, Italy
Web
KEYWORDS: THEORETICAL MOLECULAR DESCRIPTORS; GENETIC ALGORITHMS; SPLITTING; SOIL SORPTION COEFFICIENT; K-OC; QSAR; OECD PRINCIPLES; SOIL SORPTION COEFFICIENTS; ELECTROTOPOLOGICAL STATE INDEXES; PERSISTENT ORGANIC POLLUTANTS; NEURAL-NETWORKS; QSAR MODELS; QUANTITATIVE STRUCTURE; TOPOLOGICAL INDEXES; CHEMICAL-STRUCTURE; DIVERSE SET; SELECTION.