For researchers
We bring you hypotheses worth your bench time. Never noise.
Your time is the scarcest thing in rare disease, and your inbox is filling with unverified, AI-generated hunches. MISHA is a filter, not another sender. What reaches you has survived the computation, with the work shown, ready to test or reject fast.
The bottleneck is your attention, and it is under attack
Language models made it trivial to generate a plausible-sounding hypothesis. Most are wrong in ways only a specialist can see, so every unfiltered message costs the one resource that does not scale. If the field floods you, you stop reading, and rare disease cannot afford that.
The bridge, and the guard on the bridge
MISHA sits between the families who carry a case and the labs that can test it. We carry the hypothesis to you, and we guard your door. Nothing reaches you until it clears our own bar: computation is a lever, not an oracle, and everything gets checked before it moves.
Modeled, not guessed
Structure prediction, variant interpretation, and druggable-pocket analysis run first. If the computation kills the idea, it never leaves our desk.
Grounded in the literature
Every claim is checked against real, measured sources, not the model's confident summary of them. We catch the hallucination so you do not have to.
The work shown
You get the reasoning, the failed branches, and the data. You can reject it in minutes if it is wrong. That is the point.
The bar is already proven
The first packet went, unannounced, to a leading hearing-genetics lab at Harvard and to researchers in Shanghai. They engaged seriously, not because of a title, but because the analysis was worth their time. The same pipeline has merged fixes into RDKit, SageMath, and the community Lean formalization pool, and has changes under review in the tools clinical labs use to genotype STRC. That is the standard every hypothesis clears before it reaches a bench through us.
Who we are looking for
If you can test ideas, in silico or at the bench, and would give a little time to rigorously prepared hypotheses, we want to know you.
In-silico specialists
Structural biology, variant interpretation, ML for biology, protein design. Help us sharpen hypotheses before they cost anyone a pipette.
Clinicians and geneticists
You see the patients and the phenotypes. Tell us which computational leads are worth chasing and which are noise.
Wet-lab groups
When a hypothesis has cleared the computation, you are who tests it. We bring you pre-vetted, fundable leads, not homework.
How a hypothesis reaches you
A family brings a case
A parent, often a family ambassador, brings a gene, a phenotype, and urgency.
We build and verify
The pipeline ranks hypotheses, models structures, checks the literature, and formalizes what it can. Most ideas die here. That is the job.
You receive the survivors
A short, honest, wet-lab-ready packet: the hypothesis, the evidence, the failed branches, and why it is worth your time.
If you have the expertise, we will not waste it.
Tell me your field and how much time you could give. I answer every email myself, and I will only send you work that has earned it.