QSAR techniques are commonly employed for the understanding of protein chromatographic behavior. These techniques utilize knowledge-based descriptors that arise from the protein’s structure and physiochemical properties. Knowledge-based descriptors rely on clever ways of quantifying the protein’s surface properties, like using software aided potential maps, measures of hydrophobicity/hydrophilicity, molecular shapes and partial charges. A new technique of descriptor calculation is the use of powerful neural networks to understand structure-based activity. A variational autoencoder network can be used to learn protein surface properties which is independent of pre-existing knowledge. A variational autoencoder is trained to efficiently reconstruct input data by passing it through an information bottleneck. The latent space of a variational autoencoder encodes information of the input in a compressed format in the form of low dimensional vectors. These vectors can be extracted and used as descriptors which accurately represent the input data. In this work, a variational autoencoder is trained to reconstruct protein data. When the autoencoder is jointly trained with a property prediction network such as retention time predictor network, the latent space of an autoencoder encodes meaningful information in the form of vectors. These latent space vectors can be exploited as meaningful descriptors which can accurately represent protein’s structural and biophysical properties. This work demonstrates the use of data driven descriptors to perform QSAR modelling and presents the opportunity to use artificial intelligence to build descriptors for property prediction.

Using AI based descriptors to model protein chromatography

Macromolecular Chemistry: The Second Century

[Author Benefits](https://scimeetings.acs.org/?utm_source=pubs_content_marketing&utm_medium=website&utm_campaign=0320_MCF_NPI_Launch_Spring_Homepage&ref=pubs_content_marketing)

[How to use SciMeetings](https://storage.googleapis.com/dl.morressier.com/acs/SciMeetings-ACSSpring2020-HowToGuide_M3020-Final.pdf)

[Frequently Asked Questions](https://storage.googleapis.com/dl.morressier.com/acs/SciMtng_ACSNMSp2020_FAQ_M3120_Final.pdf)

ACS Spring 2020 National Meeting & Expo

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.