restructuring library as a python module

Author: decarboxy

README.md

@@ -92,6 +92,8 @@ conda activate SE3nv
 cd env/SE3Transformer
 pip install --no-cache-dir -r requirements.txt
 python setup.py install
+cd ../.. # change into the root directory of the repository
+pip install -e . # install the rfdiffusion module from the root of the repository
 ```
 Anytime you run diffusion you should be sure to activate this conda environment by running the following command:
 ```

config/inference/base.yaml

@@ -19,6 +19,8 @@ inference:
   final_step: 1
   deterministic: False
   trb_save_ckpt_path: null
+  schedule_directory_path: null
+  model_directory_path: null
 
 contigmap:
   contigs: null

examples/design_cyclic_oligos.sh

@@ -7,7 +7,7 @@
 # We decay this potential with quadratic form, so that it is applied more strongly initially
 # We specify a total length of 480aa, so each chain is 80 residues long
 
-python ../run_inference.py \
+python ../scripts/run_inference.py \
 --config-name=symmetry \
 inference.symmetry="C6" \
 inference.num_designs=10 \

examples/design_dihedral_oligos.sh

@@ -7,7 +7,7 @@
 # We decay this potential with quadratic form, so that it is applied more strongly initially
 # We specify a total length of 320aa, so each chain is 80 residues long
 
-python ../run_inference.py \
+python ../scripts/run_inference.py \
 --config-name=symmetry \
 inference.symmetry="D2" \
 inference.num_designs=10 \

examples/design_enzyme.sh

@@ -13,4 +13,4 @@
 # We specify the identity of the substrate molecule (from which to apply the potential)
 # We then specify the we want to use the model weights fine tuned to scaffold small motifs (i.e. three single residues)
 
-../run_inference.py inference.output_prefix=example_outputs/design_enzyme inference.input_pdb=input_pdbs/5an7.pdb 'contigmap.contigs=[10-100/A1083-1083/10-100/A1051-1051/10-100/A1180-1180/10-100]' potentials.guide_scale=1 'potentials.guiding_potentials=["type:substrate_contacts,s:1,r_0:8,rep_r_0:5.0,rep_s:2,rep_r_min:1"]' potentials.substrate=LLK inference.ckpt_override_path=../models/ActiveSite_ckpt.pt
+../scripts/run_inference.py inference.output_prefix=example_outputs/design_enzyme inference.input_pdb=input_pdbs/5an7.pdb 'contigmap.contigs=[10-100/A1083-1083/10-100/A1051-1051/10-100/A1180-1180/10-100]' potentials.guide_scale=1 'potentials.guiding_potentials=["type:substrate_contacts,s:1,r_0:8,rep_r_0:5.0,rep_s:2,rep_r_min:1"]' potentials.substrate=LLK inference.ckpt_override_path=../models/ActiveSite_ckpt.pt

examples/design_motifscaffolding.sh

@@ -8,4 +8,4 @@
 #   - 10-40 residues (randomly sampled)
 # We generate 10 designs
 
-../run_inference.py inference.output_prefix=example_outputs/design_motifscaffolding inference.input_pdb=input_pdbs/5TPN.pdb 'contigmap.contigs=[10-40/A163-181/10-40]' inference.num_designs=10
+../scripts/run_inference.py inference.output_prefix=example_outputs/design_motifscaffolding inference.input_pdb=input_pdbs/5TPN.pdb 'contigmap.contigs=[10-40/A163-181/10-40]' inference.num_designs=10

examples/design_motifscaffolding_inpaintseq.sh

@@ -12,4 +12,4 @@
 # We generate 10 designs
 # We then specify that residues 163-168 (inclusive), 170-171 (inclusive) and 179 (inclusive) on the A chain of the input, should be masked in the input
 
-../run_inference.py inference.output_prefix=example_outputs/design_motifscaffolding_inpaintseq inference.input_pdb=input_pdbs/5TPN.pdb 'contigmap.contigs=[10-40/A163-181/10-40]' inference.num_designs=10 'contigmap.inpaint_seq=[A163-168/A170-171/A179]'
+../scripts/run_inference.py inference.output_prefix=example_outputs/design_motifscaffolding_inpaintseq inference.input_pdb=input_pdbs/5TPN.pdb 'contigmap.contigs=[10-40/A163-181/10-40]' inference.num_designs=10 'contigmap.inpaint_seq=[A163-168/A170-171/A179]'

examples/design_nickel.sh

@@ -17,4 +17,4 @@
 
 ckpt='../models/Base_epoch8_ckpt.pt'
 
-python ../run_inference.py inference.symmetry="C4" inference.num_designs=15 inference.output_prefix=example_outputs/design_nickel 'potentials.guiding_potentials=["type:olig_contacts,weight_intra:1,weight_inter:0.06"]' potentials.olig_intra_all=True potentials.olig_inter_all=True potentials.guide_scale=2 potentials.guide_decay="quadratic" inference.input_pdb=input_pdbs/nickel_symmetric_motif.pdb 'contigmap.contigs=[50/A2-4/50/0 50/A7-9/50/0 50/A12-14/50/0 50/A17-19/50/0]' inference.ckpt_override_path=$ckpt 
+python ../scripts/run_inference.py inference.symmetry="C4" inference.num_designs=15 inference.output_prefix=example_outputs/design_nickel 'potentials.guiding_potentials=["type:olig_contacts,weight_intra:1,weight_inter:0.06"]' potentials.olig_intra_all=True potentials.olig_inter_all=True potentials.guide_scale=2 potentials.guide_decay="quadratic" inference.input_pdb=input_pdbs/nickel_symmetric_motif.pdb 'contigmap.contigs=[50/A2-4/50/0 50/A7-9/50/0 50/A12-14/50/0 50/A17-19/50/0]' inference.ckpt_override_path=$ckpt

examples/design_partialdiffusion.sh

@@ -7,4 +7,4 @@
 # But, in either case, the contig length must sum to the length of the input pdb file
 # We generate 10 designs, and noise and denoise 10 steps (20% of the full trajectory)
 
-../run_inference.py inference.output_prefix=example_outputs/design_partialdiffusion inference.input_pdb=input_pdbs/2KL8.pdb 'contigmap.contigs=[79-79]' inference.num_designs=10 diffuser.partial_T=10
+../scripts/run_inference.py inference.output_prefix=example_outputs/design_partialdiffusion inference.input_pdb=input_pdbs/2KL8.pdb 'contigmap.contigs=[79-79]' inference.num_designs=10 diffuser.partial_T=10

examples/design_partialdiffusion_withseq.sh

@@ -11,4 +11,4 @@
 # This contig will lead to the whole input pdb being noise by 10 steps (partial_T=10)
 # However, we provide the sequence of the peptide (the last 20 residues in the contig), with provide_seq=[172-205]. This is 0-indexed
 
-../run_inference.py inference.output_prefix=example_outputs/design_partialdiffusion_peptidewithsequence inference.input_pdb=input_pdbs/peptide_complex_ideal_helix.pdb 'contigmap.contigs=["172-172/0 34-34"]' diffuser.partial_T=10 inference.num_designs=10 'contigmap.provide_seq=[172-205]'
+../scripts/run_inference.py inference.output_prefix=example_outputs/design_partialdiffusion_peptidewithsequence inference.input_pdb=input_pdbs/peptide_complex_ideal_helix.pdb 'contigmap.contigs=["172-172/0 34-34"]' diffuser.partial_T=10 inference.num_designs=10 'contigmap.provide_seq=[172-205]'