Dear scientists
Nano-QSAR methods are relatively new branch of nanotoxicology. There are so many differences and a little similarity when compare to classical QSAR methods. I want to ask that which parameters can apply in nano-QSAR methods. I know that size, surface area or other spesific values can affect the nanomaterial toxicology. However, is there a standart algorithm for nano-QSAR?
I am not familiar with this particular descriptor methods; I used another method implemented (COSMOtherm software) in the past, but I think all those methods QSPR (chemical properties) or QSAR (biological properties) work in the same way. the program reads your molecular structure, calculate chemical, structural, topological,.... parameters (MM, surface area, terminal bonds, ......), compare these parameters to large data sets of known physio-chemical, toxicological,.... properties, predict the properties of your structure based on this comparison. the best practice in using these methods is to establish your own data set with limited number of response variables (MM, size, bonds, .....). collect a data set (has to be large enough) of known (experimental) toxicological properties, feed these structures into a descriptor software, use three to four similarity parameters to fit your experimental data with respect to them, change the fitting parameters until you get a reasonable accuracy, once you happy with this accuracy; use your equation to predict properties of any structure. the challenge is to reduce the number of similarity parameters into a reasonable number, if your software (nano-QSAR) gives you the percentage of how much each parameter contribute to the overall similarity (for example; size 80%, h-bonds 1%,...), this should help you choose parameters with the highest probability. hope this is helpful
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Algorithms