- Dec 20, 2003
- 13,660
- 2,692
- Country
- Germany
- Faith
- Christian
- Marital Status
- Married
I have a bunch of questions for the biologists among you.
I was chatting with a PhD in Biology today in church who is a big Pharma contractor. We were discussing a friend's illness and what could be done to find a cure. With the mapping of the human genome complete for two decades now I had hoped for more progress in finding genetic cures for diseases like the one my friend has. With the advent of AI and the ability to use AI to speed process impossible amounts of data looking for associative patterns and the relatively speedy creation of a vaccine for covid using the latest RNA techniques, I had hoped that cures for other diseases would be more readily available by now.
He mentioned the following:
Electron Microscopy:
The ability to read things like carbon bonding was not yet possible because we only have a scanning capability down to 3,2 angstroms and needed a precision of 2.7 angstroms to read the data. I googled this and saw the reference to the possibility of sub angstrom level Electron Microscopy and that models now existed to eliminate higher order aberrations from the lens images.
Is such sub angstrom microscopy useable for Big Pharma research into new drugs and cures or is this still a way off?
The AI models were not yet informed enough to be helpful.
Given the blurred vision of for example carbon bonds, no real analysis was possible anyway as the AI was operating with data that was the product of blurred vision
So has AI actually added any value if the models used are still little more than guesswork about what precise vision might actually look like? If you still have to go through an exhaustive and expensive level of physical testing of all the possibilities to see if they actually work in practice and if you cannot really predict if any of them actually work until you do?
Impacts on the Holistic System
If you switched off some ion receptors then that would have only a localized effect, with others it might have a body-wide impact. The mapping of what does what was not yet detailed and informed enough to understand how changes in one part of the system might affect changes in another.
He suggested that the statistical pattern and identification of which parts of the genetic sequence were different in people who had the disease compared to those who did not might already be available. But often the pattern might be 150 up and 150 down and a protein would be required for each transformation. If you implemented a protein to push one switch up or down no one had any idea what impacts that might have on the rest of the organism. Many proteins might be rendered nonfunctional by the changes in another.
So is all this stuff about how AI is accelerating research just baloney? The only reason that covid was done so fast was the level of resources allocated to it.
Some conceptual questions
My understanding is that the blueprint of our genome is held in the DNA. Can this be edited directly by CRISPR? How does that work?
The action-man process here is RNA and that can be used to create proteins that can lead the body to adapt. This was used by the covid vaccines and the side effects measured over billions of people did not seem to be too averse. Were we just lucky in this case of covid or was the level of resources and man-hours of testing devoted to this research the key reason why there were not any major issues or side effects?
Does the AI always look for associative patterns? Surely there are unique and entirely localized sequences also that have no analogy but have an enormous impact? How could these unique factors be mapped that cannot be revealed by analogous patterns and may not be even visible given the electron microscopes used?
I was chatting with a PhD in Biology today in church who is a big Pharma contractor. We were discussing a friend's illness and what could be done to find a cure. With the mapping of the human genome complete for two decades now I had hoped for more progress in finding genetic cures for diseases like the one my friend has. With the advent of AI and the ability to use AI to speed process impossible amounts of data looking for associative patterns and the relatively speedy creation of a vaccine for covid using the latest RNA techniques, I had hoped that cures for other diseases would be more readily available by now.
He mentioned the following:
Electron Microscopy:
The ability to read things like carbon bonding was not yet possible because we only have a scanning capability down to 3,2 angstroms and needed a precision of 2.7 angstroms to read the data. I googled this and saw the reference to the possibility of sub angstrom level Electron Microscopy and that models now existed to eliminate higher order aberrations from the lens images.
Is such sub angstrom microscopy useable for Big Pharma research into new drugs and cures or is this still a way off?
The AI models were not yet informed enough to be helpful.
Given the blurred vision of for example carbon bonds, no real analysis was possible anyway as the AI was operating with data that was the product of blurred vision
So has AI actually added any value if the models used are still little more than guesswork about what precise vision might actually look like? If you still have to go through an exhaustive and expensive level of physical testing of all the possibilities to see if they actually work in practice and if you cannot really predict if any of them actually work until you do?
Impacts on the Holistic System
If you switched off some ion receptors then that would have only a localized effect, with others it might have a body-wide impact. The mapping of what does what was not yet detailed and informed enough to understand how changes in one part of the system might affect changes in another.
He suggested that the statistical pattern and identification of which parts of the genetic sequence were different in people who had the disease compared to those who did not might already be available. But often the pattern might be 150 up and 150 down and a protein would be required for each transformation. If you implemented a protein to push one switch up or down no one had any idea what impacts that might have on the rest of the organism. Many proteins might be rendered nonfunctional by the changes in another.
So is all this stuff about how AI is accelerating research just baloney? The only reason that covid was done so fast was the level of resources allocated to it.
Some conceptual questions
My understanding is that the blueprint of our genome is held in the DNA. Can this be edited directly by CRISPR? How does that work?
The action-man process here is RNA and that can be used to create proteins that can lead the body to adapt. This was used by the covid vaccines and the side effects measured over billions of people did not seem to be too averse. Were we just lucky in this case of covid or was the level of resources and man-hours of testing devoted to this research the key reason why there were not any major issues or side effects?
Does the AI always look for associative patterns? Surely there are unique and entirely localized sequences also that have no analogy but have an enormous impact? How could these unique factors be mapped that cannot be revealed by analogous patterns and may not be even visible given the electron microscopes used?