When I am old enough to forget people’s names, I can’t even recognize the structure of a compound. I wonder what the brains of people who take the national pharmacist certification are like? Is it fun to remember such things?
We talked about this yesterday in a ZOOM meeting.
I hadn’t seen the guy I was talking to in about 20 years. I heard that she lives in Zurich, and I guess it is possible to do this now that the concept of time and distance has disappeared. It was a lot of fun.
She is also going to teach Data Science at Daiichi Pharmaceutical University.
Data science at the First Pharmaceutical University does not aim to qualify students to become pharmacists. I’m sorry to say this, but if students have to pass an exam that relies only on their memory, there is no way AI can do it. It’s common knowledge to us, but the sponsor professor said, “People in the pharmaceutical sciences think that computers are not yet capable of recognizing topological graphics such as drug molecules.
Currently, there is a project to create a database of all the pictures of molecular structures and names of compounds in pdfs on the Internet.
It’ s just a matter of time before the big data comes along.
Saying, “The compounds in medicine are so deep that it will be a long time before a computer can make medicine” is as meaningless as saying, “It will be another 30 years before a computer can beat a human at Go or Shogi.
At pirika, I have been using the SMILES formula for molecular structures for about 20 years now. This SMILES is somewhat easy to understand for both computers and humans. (INCHE is a bit more difficult for humans.)
To test this, I collected the structural formulae of SMILES for Alzheimer’s drugs.
I found cholinesterase inhibitors, ChE inhibitors, Aβ reducers, alpha4beta2 nicotinic receptors, alpha7 nicotinic receptors, etc. I put together the structures of the SMILES. I think pharmacists need to have a brain that can look at these SMILES and think of what kind of structure it is, and what kind of drug it is used for. （I can’t do it.)
***
But the computer will draw a two-dimensional molecule of whatever structure it is.
Let’s press the LOOP button. It will show you each structure in order.
In other words, even though it is difficult to recognize handwritten or printed out two-dimensional structures, the molecules themselves are properly recognized.
And if you use Y-MB in HSPiP, you can calculate various physical properties from SMILES.
For example, it calculates Hansen Solubility Parameters (HSP). (If you prefer logKow, that’s fine.)
For the solubility parameter, we can use the principle of “like dissolves like”. Let’s see.
Leaving aside the issue of the size of the molecule, what kind of solubility does the drug for Alzheimer’s have? We can see.
If we consider HSP as a 3-dimensional vector and plot it in Hansen space, it will look like the above.
You can drag the mouse to rotate, shift key and drag to zoom in/out, Option key and drag to translate.
The blue sphere shows the general solvent. Click on the blue sphere and the name of the solvent will be displayed.
The red sphere shows the Alzheimer’s drugs. Some drugs are located very close to each other. Which drug is it and what solvent is near the drug? The drug will want to be around the same environment as the solvent.
The results of molecular orbital methods, such as HOMO, LUMO, and charge, are also important information for data science, and the calculation of Memantine is as follows.
People who can use this kind of data are sought after everywhere in the world.
Do you want to become a pharmacist with a national certification that is only valid in Japan, or do you want to become a data scientist in the pharmaceutical field that is valid worldwide?
No matter how hard humans try, there is nothing humans can do about Adducanumab, C6472H10034N1742O2014S46 (protein part, four strands), and drugs are getting bigger and bigger.
I wonder if it’s time for a drug that can help me manage my dubious memory or a drug that can dramatically improve the memory of young people.
Since neither of them seems to work for the time being, I’ll rely on AI-Assist. Looking at these interactive pages, you may think that you are being tricked by a magician.
But if you want to learn these techniques, which are becoming more and more common outside of Japan (and in some cases, recurrent and working people), I look forward to seeing you at Canvas.
Which talent do you think pharmaceutical companies want?
Please leave your comments.