What data is in the QsarDB repository?
The QsarDB repository contains models that represent quantitative and qualitative relationships between chemical structures and their responses in different biological and chemical systems. A variety of relationships have been represented in the repository:
- chemico-biological activity (QSAR),
- physicochemical properties (QSPR),
- activity-activity (QSAAR),
- toxicity (QSTR),
- metabolism (QSMR),
- reactivity (QSRR),
- retention (QSRR),
- permeability (QSPR),
- pharmacokinetics (QSPR),
- bioavailability (QSBR),
- binding (QSBR),
- etc.
The QsarDB repository collets information about QSAR models, their mathematical representation and full data used for the development of model. All this data is organized into QDB archive based on scientific paper that is either published or accepted for the publication, or have other means providing scientifically sound citable description how they have been developed.
In practice, most of the statistical and mathematical algorithms of classification and regression models can be expressed as QDB archives. The examples include:
- Decision trees
- Consensus models
- k-nearest neighbors
- Logistic regression
- Neural networks
- Random forests
- Regression models
- Support vector machines
- etc.
In summary, the QsarDB is a systematically organized collection of open QDB archives in the content-aware repository.