4 ms·

Sorry is this is a lame Q: Could you take the sequences and python up a blender script that will model the thing? Also, on solutions-mixer- can I use this to

Sorry is this is a lame Q: Could you take the sequences and python up a blender script that will model the thing? Also, on solutions-mixer- can I use this to map proteins to feed to stem cells to get them to present in a certain way?

the__alchemist 2y ago
Not a lame question at all. For modeling, are you referring to protein products, or something different? If so, the answer is yes for known proteins; it includes a workflow to download structure coordinates from PDB based on an amino-acid sequence query. You can currently click a button to download these coordinates, and another to open a browser with the PDB sequence viewer open to it. It would be possible, as you said, to add functionality to open a script using Python/Blender or similar to view these, or even integrate 3D graphics directly in the program. Drawing color-coded molecules would be straightforward, but cartoon drawings etc would take more work. Modelling unknown/custom proteins would presumably require something like AlphaFold integration. The solutions mixer currently is just for molarity calculations. Like, you are making a buffer that contains 300mM of this, 100mM of that. It will tell you how many grams or mL to add of each reagent. That's a great idea re more practical workflows like presenting in stem cells! Should add more to that effect.
- samstave 2y ago
  There was a research from UCSF that I met when we were building that DOgPatch facility in Sf... She had written a program in python to pipette tiny amounts of proteins via tubular matrices valve thingy she made which allowed her to feed various proteins to stemcells via her little 3D printed tubes and then was researtching what each proteins made the stemcells behave like, what they presentd, her quote was "what do we need to ffed the stemcell to become a heart muscle." Ive alwasy been intriugued by this as I want to experiment with my own stemcells.... specifically in ocular regen. But having AI craft blender snippets has been super simple to get it to draw geometry - so a GPT that can translate between some chemistry constraints to 3D- molocule constraints - plus there are a bunch of 3D chem libraries (there was an interesting one on HN a while back that I commented on a bit..) What would be interesting would be a version of 'meta clay' (balls) -- that had attraction rules that werent meant to merge Nurbe surfaces, but rather it would know the attractions of various proteins/molocules/biochemicals such that it will know which 'covalant bonds' (I am puling that phrase out of my non-chem butt) -- work together.

the__alchemist 2y ago

Not a lame question at all. For modeling, are you referring to protein products, or something different? If so, the answer is yes for known proteins; it includes a workflow to download structure coordinates from PDB based on an amino-acid sequence query. You can currently click a button to download these coordinates, and another to open a browser with the PDB sequence viewer open to it. It would be possible, as you said, to add functionality to open a script using Python/Blender or similar to view these, or even integrate 3D graphics directly in the program. Drawing color-coded molecules would be straightforward, but cartoon drawings etc would take more work. Modelling unknown/custom proteins would presumably require something like AlphaFold integration. The solutions mixer currently is just for molarity calculations. Like, you are making a buffer that contains 300mM of this, 100mM of that. It will tell you how many grams or mL to add of each reagent. That's a great idea re more practical workflows like presenting in stem cells! Should add more to that effect.

samstave 2y ago
There was a research from UCSF that I met when we were building that DOgPatch facility in Sf... She had written a program in python to pipette tiny amounts of proteins via tubular matrices valve thingy she made which allowed her to feed various proteins to stemcells via her little 3D printed tubes and then was researtching what each proteins made the stemcells behave like, what they presentd, her quote was "what do we need to ffed the stemcell to become a heart muscle." Ive alwasy been intriugued by this as I want to experiment with my own stemcells.... specifically in ocular regen. But having AI craft blender snippets has been super simple to get it to draw geometry - so a GPT that can translate between some chemistry constraints to 3D- molocule constraints - plus there are a bunch of 3D chem libraries (there was an interesting one on HN a while back that I commented on a bit..) What would be interesting would be a version of 'meta clay' (balls) -- that had attraction rules that werent meant to merge Nurbe surfaces, but rather it would know the attractions of various proteins/molocules/biochemicals such that it will know which 'covalant bonds' (I am puling that phrase out of my non-chem butt) -- work together.