Protein deep-moth-lava

Seongjun Yang / deep-moth-lava

Type

Unknown

Ref

design_339

Collections

S[GEM x Adaptyv: RBX1 Binder Design Competition] Submission 1

AGEM x Adaptyv RBX1 Binder Competition: Main Submissions

Sequence (72 AA)

SLVEQRREEVRRRFRESYEEWSRVAERDPESLRKGREWREWGERTGPEGMGVSSWYLESGLRRERERIRRER

Computational Metrics

Experimental Metrics

Structure prediction unavailable

It might take a few hours to appear.

No experimental data

This protein hasn't been validated in the lab yet.

Design Method

This protein was designed using SYRFdiffusion + LigandMPNN + AF3 ipSAE Selection

Target: RBX1 RING domain. Backbone generation: RFdiffusion with E2-surface hotspots only; helical campaigns only. Sequence design: LigandMPNN (zinc-aware), temperature 0.2, 16 sequences per backbone, amino acid biases A:-3.0, W:+2.0, Y:+1.5, F:+1.0, I:+0.5, V:+0.3, C:-999.0. Filtering: sequence quality filters and AF2 monomer screen (pLDDT >= 85). Final selection: AF3 validation with ipSAE_d0chn as the primary ranking metric, gate-based filtering for clash-free models, disorder <= 10%, binder pTM >= 0.70, RBX1 pTM >= 0.70, and diversity capped at 5 sequences per backbone.