(no subject)
Back in the Pleistocene, I was walking through a basement corridor at MIT and passed the door of what must have been a bio lab. On the glass was taped a cartoon which was titled something like "The four ages of DNA". The first frame was a stone tablet with the letters DNA chiseled in it. "DNA exists and is important but we have no idea what it is." The second frame was the letters DNA assembled from lines which were double helixes. "We know the chemistry of DNA." The third frame was the letters DNA assembled from the girders of an Erector set. "We can splice existing genes from one organism to another." The fourth frame was in the style of a microphotograph of an integrated circuit with the various blobs arranged to look like the letters DNA. (This was back when an integrated circuit had sufficiently few transistors that you could see patterns and details in the image.) "We can build DNA that does what we want."
We were in the 3rd age then and we've been in the 3rd age since then. But we're entering the 4th age:
https://www.science.org/doi/10.1126/science.abq0225
Decoding CAR T cell phenotype using combinatorial signaling motif libraries and machine learning
They're starting to produce immune receptor proteins that specifically target the particular genetic abnormalities that certain cancers exhibit, and you can use these as drugs that target and kill the cancer cells. Apparently, it's not so difficult to get receptors that grab and are triggered by the abnormality that you're going after, but it's hard to get receptors that trigger the immune T cells in exactly the way you want. It looks like there is a "signaling part" of the receptor that's composed of a bunch of parts, "signaling motifs", but there's not a neat 1-1 correspondence between the motifs and the T cell activities that they trigger. So someone has tested what "200 of 2400 possible combinations of 13 signaling motifs" do, and programmed a machine learning model to predict with some accuracy what any combination will do.
We were in the 3rd age then and we've been in the 3rd age since then. But we're entering the 4th age:
https://www.science.org/doi/10.1126/science.abq0225
Decoding CAR T cell phenotype using combinatorial signaling motif libraries and machine learning
They're starting to produce immune receptor proteins that specifically target the particular genetic abnormalities that certain cancers exhibit, and you can use these as drugs that target and kill the cancer cells. Apparently, it's not so difficult to get receptors that grab and are triggered by the abnormality that you're going after, but it's hard to get receptors that trigger the immune T cells in exactly the way you want. It looks like there is a "signaling part" of the receptor that's composed of a bunch of parts, "signaling motifs", but there's not a neat 1-1 correspondence between the motifs and the T cell activities that they trigger. So someone has tested what "200 of 2400 possible combinations of 13 signaling motifs" do, and programmed a machine learning model to predict with some accuracy what any combination will do.