In the original paper,backbones for 1000 mini-binders were created using RFdiffusion by conditioning the model on the 3D structure of IL-7Ra. For each of the binder structures generated, 2 sequences were designed using the ProteinMPNN inverse folding model. Finally, these sequences were folded using AlphaFold2 and the designs were filtered based on several computational metrics of the model (pae_interaction < 10).
id: quick-panther-snow

IL-7Ra
Strong
4.0e-8 M
True
7.2 kDa
65
id: swift-panda-snow

IL-7Ra
Medium
5.3e-8 M
True
9.5 kDa
80
id: solid-falcon-thorn

IL-7Ra
Medium
6.4e-8 M
True
9.9 kDa
86
id: scarlet-bear-lotus

IL-7Ra
Medium
6.4e-8 M
True
10.4 kDa
88
id: scarlet-ant-granite

IL-7Ra
Medium
7.7e-8 M
True
9.8 kDa
85
id: steady-swan-onyx

IL-7Ra
Medium
9.4e-8 M
True
10.9 kDa
97
id: lunar-toad-cypress

IL-7Ra
Medium
1.2e-7 M
True
10.2 kDa
83
id: azure-bear-oak

IL-7Ra
Medium
1.3e-7 M
True
7.5 kDa
65
id: azure-eagle-maple

IL-7Ra
Medium
1.4e-7 M
True
7.1 kDa
64
id: lunar-jaguar-thorn

IL-7Ra
Medium
1.8e-7 M
True
7.3 kDa
65
id: vast-ant-crystal

IL-7Ra
Medium
2.4e-7 M
True
10.2 kDa
90
id: calm-vole-fern

IL-7Ra
Medium
2.7e-7 M
True
7.5 kDa
67
id: quick-boar-ruby

IL-7Ra
Medium
3.3e-7 M
True
6.5 kDa
54
id: dark-falcon-cedar

IL-7Ra
Medium
3.6e-7 M
True
7.9 kDa
68
id: ivory-bat-moss

IL-7Ra
Medium
8.3e-7 M
True
7.3 kDa
65
id: shy-tiger-pine

IL-7Ra
None
--
True
9.8 kDa
84
id: jade-jaguar-plume

IL-7Ra
Weak
--
True
6.8 kDa
58
id: rough-cobra-thorn

IL-7Ra
Weak
--
True
10.0 kDa
88
id: wild-crane-rose

IL-7Ra
Weak
--
True
11.0 kDa
97
id: shy-orca-sand

IL-7Ra
Weak
--
True
10.0 kDa
91
id: solid-raven-onyx

IL-7Ra
Weak
--
True
8.9 kDa
81