Anna Greka

And so there's multiple different kinds of perturbations that we do to cells, whether we're using

CRISPR or base editing to make, for example, genome-wide or genome-scale perturbations or small molecules, as we have done as well in the past.

These are all ways in which we are then using machine learning to read out the effects in images of cells that we're looking at.

So that's one way in which machine learning is used in our daily work.

Of course, because we study misshapen mangled proteins and how they are recognized by these cargo receptors, we also use alpha-fold pretty much every day in my lab.

And this has been catalytic for us as a tool because we, you know, we really are able to accelerate our discoveries in ways that were even just...

you know, three or four years ago, completely impossible.

So it's been incredible to see how, you know, the young people in my lab are just so excited to use these tools and they're becoming extremely savvy, you know, in using these tools.

Of course, this is a new generation of scientists.

And so we use AlphaFold all the time.

And this also has a lot of implications, of course, for some of the interventions that we might think about.

So, you know, where in this

you know, cargo receptor complex that we study, for example, might we be able to fit a drug that would disrupt the complex and lead the cargo trucks into the lysosome for degradation, for example.

So, you know, there's many ways in which AI can be used for all of these functions.

So I would say that if we were to organize our thinking around it

One way to think about the use of machine learning AI is around what I would call understanding biology in cells and what, you know, in sort of more kind of drug discovery terms, you would call target identification, like trying to understand the things that we might want to intervene on in order to have a benefit for disease.

So target ID is one area in which I think, you know, machine learning and AI will have a catalytic effect as they already are.

The other, of course, is in the actual, you know, development of the appropriate drugs in a rational way.

So like rational drug design is incredibly enabled by alpha fold and all these advances in terms of understanding protein structures and how to fit drugs into them of all different modalities and kinds.

And I think an area that we are not yet harnessing in my group, but I think the latter secures accelerator hopes to build on, is really patient data.