This web server provides online access to realize a full workflow of building a Structure‑Activity/Property Relationship(SAR/SPR) model. It provides a browser-based pipelining which requires no programming skills. Therefore, It is applicable for users including pharmacologists and biological scientists, not only the computational chemistry or chemoinformatics scientists. To start using it, please click the menu item on the top. If you prefer an example or get some problems please refer the 'Help' section.

Please cite: Jie Dong, Zhi-Jiang Yao, Min-Feng Zhu, Ning-Ning Wang, Ben Lu, Alex F Chen, Ai-Ping Lu, Hongyu Miao, Wen-Bin Zeng, Dong-Sheng Cao. ChemSAR: an online pipelining platform for molecular SAR modeling. Journal of Cheminformatics. 2017, 9(1): 27. [PDF]

News & Updates

  • Version 1.0(2016-5-20). The ChemSAR Version 1.0 was released.
  • Version 1.2(2016-8-10). Added 'Diverse training set split'.
  • Version 1.3(2016-9-25). Added 'Decision tree' and 'SVM' methods.
  • Version 1.4(2016-9-29). Allow to store user's file.
  • Version 2.0(2017-5-11). Added 'My Model', 'File Check', 'S indices', ''Feature importance' , 'Tree visualization' modules; added some tips at several modules; added EXPLANATIONS in 'Manual'.
  • Version 2.1(2017-6-08). Fixed bugs in 'Statistical Analysis'.(Thank Defang Fan.)
  • Version 2.2(2017-8-23). Fixed bug in 'Feature calculation'. bug: The 'nsulph' and 'noxy' are the same.(Thank Ermanno Oberrauch.)
  • Version 2.3(2017-11-02). Fixed bug in 'Prediction'. bug: 'Server Error 500'. (Thank Defang Fan.)

In silico predictive models have proved to be valuable for the optimisation of compound potency, selectivity and safety profiles in the drug discovery process. ChemSAR is an web-accessible platform that provides an easy-to-use interface for the rapid generation of quantitative Structure‑Property and Structure‑Activity classification models for small molecules (including SAR, SPR, SAM) and is aimed at both advanced and users without any programming skills. ChemSAR’s capabilities include the validation and standardisation of chemical structure representation, computation of 783 1D/2D molecular descriptors and 10 types of commonly used fingerprints for small molecules, filtering methods for feature selection, generation of predictive models (a step-by-step job submission process), as well as a helpful report generation system. Results can be visualised through high-quality, customisable plots and be downloaded as certain files.

User space

  • A file system that allows users to view files they uploaded in each step.
  • According to the tags users could reuse these files conveniently.
  • Get started >>>


  • ChemSAR provides several tools for preprocessing.
  • It mainly contains structure preprocessing for molecules and data preprocessing for the dataset.
  • Get started >>>

Modelling process

Report system

  • A specific feature of ChemSAR is that it provides a complete report generation system.
  • It retrieves result data of each calculation and transforms them into a HTML page and a PDF for users.
  • Get started >>>