boltz
boltz copied to clipboard
jwohlwend
Official repository for the Boltz biomolecular interaction models
I don't know why, but after using boltz-1 for prediction, a cyclohexane structure in my ligand turned into a benzene ring. Since I'm not very familiar with the SMILES format,...
I am experiencing a discrepancy between the ligand structure I expect based on my input SMILES and the 3D structure generated by Boltz. My SMILES string specifies an alkane chain...
>A|protein| MLGPAVLGLSLWALLHPGTGAPLCLSQQLRMKGDYVLGGLFPLGEAEEAGLRSRTRPSSPVCTRFSSNGLLWALAMKMAVEEINNKSDLLPGLRLGYDLFDTCSEPVVAMKPSLMFLAKAGSRDIAAYCNYTQYQPRVLAVIGPHSSELAMVTGKFFSFFLMPQVSYGASMELLSARETFPSFFRTVPSDRVQLTAAAELLQEFGWNWVAALGSDDEYGRQGLSIFSALAAARGICIAHEGLVPLPRADDSRLGKVQDVLHQVNQSSVQVVLLFASVHAAHALFNYSISSRLSPKVWVASEAWLTSDLVMGLPGMAQMGTVLGFLQRGAQLHEFPQYVKTHLALATDPAFCSALGEREQGLEEDVVGQRCPQCDCITLQNVSAGLNHHQTFSVYAAVYSVAQALHNTLQCNASGCPAQDPVKPWQLLENMYNLTFHVGGLPLRFDSSGNVDMEYDLKLWVWQGSVPRLHDVGRFNGSLRTERLKIRWHTSDNQKPVSRCSRQCQEGQVRRVKGFHSCCYDCVDCEAGSYRQNPDDIACTFCGQDEWSPERSTRCFRRRSRFLAWGEPAVLLLLLLLSLALGLVLAALGLFVHHRDSPLVQASGGPLACFGLVCLGLVCLSVLLFPGQPSPARCLAQQPLSHLPLTGCLSTLFLQAAEIFVESELPLSWADRLSGCLRGPWAWLVVLLAMLVEVALCTWYLVAFPPEVVTDWHMLPTEALVHCRTRSWVSFGLAHATNATLAFLCFLGTFLVRSQPGCYNRARGLTFAMLAYFITWVSFVPLLANVQVVLRPAVQMGALLLCVLGILAAFHLPRCYLLMRQPGLNTPEFFLGGGPGDAQGQNDGNTGNQGKHE >B|protein| MGPRAKTISSLFFLLWVLAEPAENSDFYLPGDYLLGGLFSLHANMKGIVHLNFLQVPMCKEYEVKVIGYNLMQAMRFAVEEINNDSSLLPGVLLGYEIVDVCYISNNVQPVLYFLAHEDNLLPIQEDYSNYISRVVAVIGPDNSESVMTVANFLSLFLLPQITYSAISDELRDKVRFPALLRTTPSADHHIEAMVQLMLHFRWNWIIVLVSSDTYGRDNGQLLGERVARRDICIAFQETLPTLQPNQNMTSEERQRLVTIVDKLQQSTARVVVVFSPDLTLYHFFNEVLRQNFTGAVWIASESWAIDPVLHNLTELRHLGTFLGITIQSVPIPGFSEFREWGPQAGPPPLSRTSQSYTCNQECDNCLNATLSFNTILRLSGERVVYSVYSAVYAVAHALHSLLGCDKSTCTKRVVYPWQLLEEIWKVNFTLLDHQIFFDPQGDVALHLEIVQWQWDRSQNPFQSVASYYPLQRQLKNIQDISWHTINNTIPMSMCSKRCQSGQKKKPVGIHVCCFECIDCLPGTFLNHTEDEYECQACPNNEWSYQSETSCFKRQLVFLEWHEAPTIAVALLAALGFLSTLAILVIFWRHFQTPIVRSAGGPMCFLMLTLLLVAYMVVPVYVGPPKVSTCLCRQALFPLCFTICISCIAVRSFQIVCAFKMASRFPRAYSYWVRYQGPYVSMAFITVLKMVIVVIGMLATGLSPTTRTDPDDPKITIVSCNPNYRNSLLFNTSLDLLLSVVGFSFAYMGKELPTNYNEAKFITLSMTFYFTSSVSLCTFMSAYSGVLVTIVDLLVTVLNLLAISLGYFGPKCYMILFYPERNTPAYFNSMIQGYTMRRD ------------------------- COMMANDD LINE: boltz predict "$fasta_file" \ --out_dir "$output_binder_dir" \ --cache "$CACHE_DIR" \ --accelerator "$ACCELERATOR" \ --recycling_steps "$RECYCLING_STEPS" \ --sampling_steps "$SAMPLING_STEPS" \ --diffusion_samples "$DIFFUSION_SAMPLES" \ --step_scale...
Hi, thanks for the wonderful implementation and repo! I was experimenting with Boltz1 on a kinase: ``` version: 1 # Optional, defaults to 1 sequences: - protein: id: ABL sequence:...
Hi, I have set the accelerator to cpu (`boltz predict examples/ligand.fasta --accelerator cpu`), but the following erros still happened: Running on CPU, this will be slow. Consider using a GPU....
I believe the reshape on line 161 in `smooth_lddt_loss` is incorrect and will misalign `mask` and `eps` if both batch and multiplicity are larger than 1. The reason is that...
This is an exercpt of my yaml setup ``` sequences: - protein: id: seq_2 msa: seekgene_msa/N42_cdr3/seq_2.a3m sequence: CSVGTGDFGEQYF - protein: id: seq_4 msa: seekgene_msa/N42_cdr3/seq_4.a3m sequence: CATSFSGPEQFF - protein: id: seq_6...
Highly appreciated the great work! I have a question of MSAs for multimers. It seems that the a3m file from colabfold_search contains both paired and unpaired MSAs, to run Boltz...
We noticed that the ligand bond order is not retained in the boltz PDB output, as noted in a related issue for the mmCIF output (https://github.com/jwohlwend/boltz/issues/49). By changing the PDB...
Dear Developer, I have two workstations. I set up the calculation on the first computer and transferred the data and conditions to the second computer(Generated requirements.txt at the first computer...