Hello, I am Willing to perform a QSAR analysis. Meanwhile, I have only 12 compounds in my library. On analysis, I found a good r2 value of 0.97 and s strong correlation diagram between experimental pIC50 and predicted pIC50. The problem is there's only one basis set for this number of compounds. Is it possible to include this analysis?
yes u can do QSAR with 12 compounds, but it's better with more compounds , so u can get better result and .
Yes, it is possible to perform QSAR analysis for 12 compounds, but the results may not be as robust or reliable as those obtained with a larger dataset. While you have observed a strong correlation (r2 = 0.97) between experimental and predicted pIC50 values, it is essential to note that this may be due to overfitting, given the small sample size.
Overfitting occurs when a model captures noise in the data and learns complex relationships that do not generalize well to new, unseen data. With a small dataset of 12 compounds, it is more likely that your QSAR model is overfitting, which could lead to poor predictive performance on new compounds.
To improve the reliability of your analysis, consider the following suggestions:
Yes, it is possible to perform QSAR analysis for a dataset of 12 compounds. But, the statistical significance of the results may be limited due to the small number of compounds. It is generally advised to have a bigger dataset, ideally with at least 50-100 compounds, to improve the reliability and robustness of QSAR models.
In addition, the dataset needs to be diverse and reliable in terms of the chemical space. A small dataset might not include all of the structural and physicochemical characteristics required to build a trustworthy QSAR model. Therefore, it is crucial to carefully select the compounds in the dataset.
Thank you Kaushik Sarkar and Alireza Mohebbi for your valuable suggestiona
- Badges
- Science topic
- Similar topics
- Engineering
- Chemical Engineering
- Membranes