Pro-1 is a fully open source reasoning model (8b and 70b) trained using GRPO towards a physics based reward function for protein stability. It takes in a protein sequence + text description of the protein + past mutagenesis results, reasons over the information given, and proposes modifications to improve the stability of the given sequence.
id: amber-goat-onyx

FGF R1
Medium
9.6e-8 M
True
17.5 kDa
155
id: steady-moth-lotus

FGF R1
Medium
5.9e-8 M
True
17.5 kDa
155
id: brisk-deer-flint

FGF R1
Strong
3.2e-8 M
True
17.4 kDa
155
id: radiant-shark-iron

FGF R1
Strong
3.4e-8 M
True
34.9 kDa
311
id: bright-cobra-leaf

FGF R1
Strong
2.1e-8 M
True
17.5 kDa
155
id: gentle-cobra-ice

FGF R1
Strong
1.5e-8 M
True
17.5 kDa
155
id: rough-lion-opal

FGF R1
Strong
8.7e-9 M
True
17.5 kDa
155
id: swift-cobra-birch

FGF R1
None
81.85
True
17.4 kDa
155
id: scarlet-raven-snow

FGF R1
None
88.74
True
34.1 kDa
305
id: amber-ram-thorn

FGF R1
None
81.42
True
17.4 kDa
155
id: small-otter-sand

FGF R1
None
81.51
False
17.8 kDa
156
id: rough-otter-birch

FGF R1
None
87.09
False
17.4 kDa
155
id: small-deer-rose

FGF R1
None
87.13
True
17.5 kDa
155
id: amber-cobra-lotus

FGF R1
None
87.03
True
17.5 kDa
155
id: steady-gecko-onyx

FGF R1
None
85.63
True
17.4 kDa
155
id: lunar-yak-flint

FGF R1
None
85.41
True
17.5 kDa
156
id: noble-zebra-ash

FGF R1
None
87.26
True
17.3 kDa
155
id: hollow-ox-topaz

FGF R1
None
83.00
True
17.4 kDa
155
id: lunar-gecko-opal

FGF R1
None
86.93
False
17.2 kDa
154
id: frozen-bison-pine

FGF R1
None
87.55
True
17.4 kDa
155