This is a remark from the Editor-in-Chief which of the submitted QSAR/QSPR papers will be rejected in the future. It is based on GLP (good laboratory practice) for computational chemistry and the ACS Ethical Guidelines, which were just unappreciated by some of the former journal editors and reviewers in QSAR research journals. This includes some of the points:

• The QSAR study or methodology or theory must be truly new :-)
• All used data, molecules, descriptors must be published, to permit the author's peers to repeat the work. Please see Reverse engineering chemical structures from molecular descriptors: how many solutions?
• QSAR/QSPR model have been properly validated using data which was not in the training set.
• 3D-QSAR studies which overlap with structure-based design methods are encouraged.
• Papers which facilitate a mechanistic understanding of adsorption, distribution, metabolism, excretion, and toxicity (ADMET) properties are warmly welcomed.
• QSAR/QSPR methods that interface with chemo-, bioinformatics and data mining methods are heartily encouraged.
• Only QSAR analyses that bring new insights on the mechanism of activity are encouraged.
• QSAR/QSPR studies for new, novel endpoints, both biological and physical using new experimental data are encouraged.
• Not allowed or specifically discouraged are QSAR and QSPR modeling for data sets which (1) have already been extensively modeled, (2) model development featuring high ratios of descriptors to data points, (3) reports of new descriptors without clear evidence for their superiority in QSAR/QSPR modeling to existing or commonly used models.

QSAR research paid by drug companies usually forbids to publish data about molecules in an early state, hence when patents are not filed yet. As the drug pipeline or whole development of a new drug may last 5-30 years it may be important to publish QSAR or QSPR studies during this time. Protecting or covering or hiding the true molecule structure maybe an important issue.

If multiple molecular descriptors are published (like in a good QSAR/QSPR study) the true molecule structure can be easily found out aka reverse engineered. In this study more than 76% out of a test sample from 500,000 structures from the PubChem database could be potentially reverse engineered. The authors conclude that sharing useful information about chemical compounds without revealing their structures is a challenging problem. However they did not have a systematic technique to reverse engineer all molecular structures. (Remark: This is a pure computer speed problem which may be solved in the post-metabolomics era).

The authors suggest using molecular fragments instead of topological indices for sharing information on QSAR studies.