Target Atlas
Profiles for the Dark Genome
A growing, open library of computational profiles for proteins the field has left dark — no solved structure, no known ligand, thin literature. Each resolves a sequence into a topology or fold, a disorder map, modification sites, and a predicted pocket, with a validation plan.
Have a dark or under-characterised target? Nominate it — membrane or soluble.
GPR160
Q9UJ42 · 338 aa · PDB: none
The dark oncology / pain orphan with a high-priority pocket hypothesis. No solved structure, no confirmed ligand, contested CARTp biology — resolved into a topology, a residue-level orthosteric-pocket hypothesis, and a validation plan.
GPR149
Q86SP6 · 731 aa · PDB: none
The oversized, disorder-dominated orphan where construct design is the product — an unusually large (731-aa), ~65%-disordered orphan GPCR mapped into truncation boundaries, a pocket hypothesis, and a validation plan.
GPR85
P60893 · 370 aa · PDB: none
The most conserved vertebrate GPCR, still an orphan — a medium-confidence, lipid-adjacent pocket (not forced into a high-confidence claim), with topology, PTM and validation guidance.
SLC38A11
Q08AI6 · 406 aa · PDB: none
The "putative" SLC38 member, deorphaned in silico — a dark, structureless multi-pass transporter where Astra confirms the transporter identity (p = 1.00) and predicts the substrate-binding cavity.
TMC7
Q7Z402 · 723 aa · PDB: none
The PIEZO2-tuning channel no one has solved — a structureless multi-pass TMC-family protein where Astra confirms the topology and predicts the pore-lining cavity, for a channel that sets the gain on touch and mechanical pain.
ADPRM
Q3LIE5 · 342 aa · PDB: none
A confirmed enzyme with an unsolved human active site — a soluble Mn-dependent hydrolase where Astra confirms the enzyme class and predicts a di-metal active-site pocket that lands on its catalytic residues.
MDH1B
Q5I0G3 · 518 aa · PDB: none
A soluble "putative" malate dehydrogenase whose activity has never been shown — Astra confirms it is a real oxidoreductase and flags a high-confidence pocket in its unusual N-terminal extension.
OVCA2
Q8WZ82 · 227 aa · PDB: none
A soluble esterase from an ovarian-cancer deletion locus — Astra recovers its hydrolase identity from sequence and lands the predicted pocket on the Ser–His–Asp catalytic triad, for a protein with no solved structure and no known substrate.
MINDY4
Q4G0A6 · 757 aa · PDB: none
A predicted function, an active-site pocket that lands on the catalytic dyad, and a modification map — for a soluble deubiquitinase with no solved structure and unconfirmed activity.
SCYL3
Q8IZE3 · 742 aa · PDB: none
A predicted function call, a degenerate kinase pocket, and a modification map — for a soluble pseudokinase with no solved structure and no known substrate.
TMEM43
Q9BTV4 · 400 aa · PDB: none
LUMA — behind a lethal cardiomyopathy and two more Mendelian diseases; a structureless inner-nuclear-membrane protein resolved into a topology, a disease-variant structural map, and a predicted pocket.
ABI3
Q9P2A4 · 366 aa · PDB: none
A top-tier Alzheimer's genetic hit whose mechanism is unknown — a structureless, half-disordered adaptor resolved into a domain map, a construct-design plan, and an interaction pocket on its SH3 domain.
FAM222A
Q5U5X8 · 452 aa · PDB: none
"Aggregatin" — an Aβ-nucleating protein with no structure that AlphaFold resolves poorly; Astra maps its order/disorder and flags a tractable, high-confidence pocket distinct from its Aβ-binding region.
SAMD9L
Q8IVG5 · 1,584 aa · PDB: none
Three inherited diseases, one structureless giant — a huge multidomain effector behind ataxia-pancytopenia, monosomy-7 myelodysplasia and SCA49; Astra maps its architecture, places disease variants in context, and predicts an effector-core pocket.
WDR89
Q96FK6 · 387 aa · PDB: none
The conserved-but-dark propeller, newly tied to genome maintenance — a predicted seven-bladed β-propeller with no solved structure, resolved into a fold, a protein-interaction pocket, and a validation plan.
Turn a Dark Target Into an Experiment-Ready Plan
Send a UniProt ID and Orbion returns a characterisation preview like the ones above — topology or fold, disorder, PTMs, a pocket or active-site hypothesis, and recommended validation experiments. New targets go up every week.