There is a reason why toxicologists love chemical structures. Because structures tell stories. And there is this cool model with a not-so-cool name that helps translate chemical goop into interesting narratives. It is called Quantitative Structure-Activity Relationship or QSAR (I warned you). It is the correlation between the 3D structure of a molecule and its biological activity.
Assumptions of QSAR
The underlying assumption is that the structure of a compound (i.e., its atoms and bonds) determines its properties. Hence, similarly-structured compounds have similar activities (or toxicities).
In toxicology, QSAR is often used to predict if:
1. A compound’s structure can cause a specific toxic effect
2. A compound’s structure can predict the potency of the toxic effect
To answer this, a QSAR model has to be developed first. This requires information on the structure and activity of several compounds that fulfill a similar criteria or are from the same group. Statistical analyses are carried out to link the compounds’ structure to their activity. If this link is found to be robust, the model can be used to predict the activity of other (untested) compounds that are in the same group.
Here are some examples that were successfully tested by QSAR:
- With hydrocarbons (compounds made of carbon and hydrogen), increasing the number of carbon atoms decreases the volatility of the compound
- Effective drugs are more lipophilic (fat-loving) than hydrophilic (water-loving) as lipophilic drugs are more likely to be absorbed by the body
- If a compound has -NH atoms attached to a ring or has a three-atom ring with oxygen in the center, it is more likely to damage DNA
Advantages & disadvantages of QSAR
QSAR can be used to predict the activities of chemicals that have not been studied in the lab. In fact the EPA has often used QSAR to predict the toxicity and exposure potential of non-pesticide chemicals before allowing release of the chemical into the environment. The information from QSAR can be also be used to conduct more efficient risk assessments (see figure below) and to create new compounds (drugs or pesticides).
But QSAR can also make false predictions, especially if the experimental data used were insufficient. In fact, the biggest limitation of QSAR is that it is only as good as the availability and quality of the ‘training’ data. Also, a structure and a property could be merely correlated, and therefore it is important to use the correct data to carry out QSAR analyses.