The recent 20th International Workshop on (Q)SAR in Environmental and Health Sciences (QSAR 2023) provided an important international stage for Drs. Ruchir Shah and Vijay Gombar to showcase groundbreaking advancements in Sciome’s state-of-the-art cheminformatics tools. Hosted in Europe, this prestigious conference highlighted key innovations and fostered discussions on evolving methodologies.
As the scientific community increasingly shifts away from traditional animal testing, effectively harnessing detailed data from New Approach Methodologies (NAMs) has become vital. Sciome is at the forefront of addressing these emerging data-analysis and interpretation challenges, presenting three significant developments: OrbiTox, Saagar, and SCiP.
OrbiTox – A Revolutionary 3D Environment for Toxicological Data Analysis
OrbiTox represents a paradigm shift in predictive toxicology platforms, integrating extensive, expertly curated datasets with hundreds of validated QSAR models. By efficiently handling complex, high-dimensional, multi-domain data, OrbiTox enables researchers to seamlessly extract meaningful toxicological insights from millions of interconnected scientific data points. The intuitive user interface provides streamlined access to a curated collection featuring 900,000 substances, 22,000 human targets, 1,500 biological pathways, over 100 test organisms, and predictive models for more than 40 Tox21 assays and bacterial mutagenicity tests. This innovative platform also offers real-time 3D visualization of data, effortlessly scalable to millions of points, distinguishing OrbiTox from existing graph-visualization solutions.
Saagar – Advancing Interpretable Predictive Models and Read-Across Protocols
Sciome introduces Saagar, an innovative suite of extensible, chemistry-aware substructures that enhance interpretable predictive models and read-across protocols. Through rigorous research and extensive testing, Saagar has demonstrated superior performance compared to publicly available fingerprint and benchmark sets such as ToxPrint and Mordred. Its interpretable and efficient chemical characterization capabilities make Saagar ideal for developing reliable in silico models and read-across approaches, providing researchers deeper and more actionable insights.
SCiP – The Next-Generation Cheminformatics Platform
Leveraging the power of Saagar, Sciome’s advanced cheminformatics platform, SCiP, significantly enhances descriptor matrix generation. SCiP creates comprehensive descriptor matrices that incorporate the 834 Saagar v1 features per input dataset, including SMILES notations and additional user-provided data. It also supports custom structural features encoded as SMARTS. Descriptor vectors generated by SCiP can be employed for QSAR model development using Recursive Partitioning (RP) within the platform or exported for use with other modeling approaches. Saagar-RP models developed within SCiP have demonstrated comparable or superior predictive performance to those built using Mordred descriptors and traditional modeling methods such as Random Forest, Support Vector Machines, and Deep Neural Networks. Additionally, Saagar-RP models provide clear, chemically-supported explanations for each prediction, greatly enhancing interpretability and practical utility.
For further information regarding our advanced cheminformatics tools, please contact us at software.support@sciome.com.
