Hi all, this is Naveen from Immunity Project YC team. The fact that we are using ML to identify epitopes that are the preferred targets on the virus for HIV controllers is key to what makes our project unique. This is a newer approach for developing a vaccine and therefore makes it "rogue."
By the way. That "rouge" approach is what would I have go through as a sysadmin if this was an computer related problem. I think you deserve to be a case study in immunology and be very very successful.
Sometimes thinking out of the box and letting go of established ways will lead to results.
Edit:
Also in computer sciences we use differentiating between two known states to find out how to change on state to the other one perfectly. In immunology do you use similar approaches?
Hey Naveen, I know that there are some epitopes that can only be recognized if there's a weird domain swap that happens, causing antibodies to be "four-pronged" instead of "two pronged". I think it's a proline substitution in the conserved neck region. Did you screen out those epitopes from your ML?
Also good luck! I'm chipping in now. I'm relaunching my own crowd-funded nonprofit research in the biomedical space (and will be partnering with crowdhoster/crowdtilt)... I'm magnifying your message. When the dust settles a little bit I'd love to be in touch with you guys.
