In Silico Olfaction

 

15.11.2022

Introducing OLaF our in-silico Olfactory Model

 

As many of you already noticed, I have been bragging about my computer model of the human olfactory receptors array. We named it  Olfactory is one of the least understood sensory modalities of the human body and other animals. Our current theories are ones forged as late as the 1990s.

 

A lot of the complexity arises from the large number of olfactory receptors and the fact that they don't have a one-to-one relationship with any specific odorant ligand. Our current theories suggest that an odorant molecule will interact with many olfactory receptors. Therefore, the unique odor sensation is determined by the unique activation pattern of the odorants across the array of receptors.

As far as I know, there is not yet a serious scientific research paper describing such a computer model for in-silicio olfaction. Still, I am sure this is coming very soon. If we could build one, some capable academic researchers would do the same. The enabling factor is the fantastic advancements we all experience in the field of Machine Learning.

As you should have noticed by now, I am an avid supporter of openness in the fragrance industry, yet as this is a part of a commercial project, I can't expose all the details. Nonetheless, feel free to ask whatever you want, and I will do my best to answer within my limitation.

Not going into too many details, our model consists of an internal representation of the 400 plus G protein-coupled receptors. We then match each one (a bit more than 4000) of the odorant molecules in our database against all of the receptors. We are using the Monte Carlo method physics simulation, coupled with state-of-the-art Neural Network technology, to create an affinity score for each of the odorants to each of the receptors.

I promise to tell you more about it soon, but until then, I feel like showing off some photos of the patterns we receive for a few of the more famous odor molecules. Finding their similarities will be left as an exercise (I will share my insights in a future post). Some of the photos represent the activation over all the receptors. In others, I focus on some receptors to emphasize the relations between different activation patterns.

Musk Ketone.png
Romandolide.png
Galaxolide.png
Ethylene Brassylate.png
Helvetolide.png
musks_activation (1).png
musks_activation_focused (1).png

To be continued...