* Update README.md
* Reverted to Geoffrey's version
* Merged with github
* Sequential implementation for batch mode
* Added new script for mmt
* Parallel implementation
* use create_db() also as a standalone script
* Multiple templates for each fasta file
* Keep templates with identical sequences for TrueMultimer
* Reduce MSA depth for each next model if run with --multimeric_mode=True
* use np.linspace() for msa depths
* c_i_f_w_t.py: Reuse functions from c_i_f.py; c_i_f.py: Remove mmt, back to main branch
* Remove temp files
* Forking.md removed
* Commented out saving multichain mask used for debugging
* Changed starting values for num_msa and num_extra_msa for TrueMultimer
* First version of json metadata file. Added __version__
* Cosmetic improvements
* New more detailed meta.json
* Save time to meta.json if mmseq2=True
* Removed accidently added colabfold script
* Update .gitmodules
* Revert file to version from main branch
* Revert files to version from main branch
* Update run_multimer_jobs.py
* Updated submodule URL
* Update to the latest AlphaFold version
* Remove version for openmm and pdbfixer
* Save only particular chain to db using biopython
* fix truncated file, not original
* metadata output is json
* no truncation of the templates by default
* Remove all lines with UNK residue from the mmcif string for templates
* Replace entity_poly_seq table by full sequence for templates
* Update packages, python 3.1
* Update alphafold to the latest commit from the main branch
* Update URL for colabfold
* Update README.md
* Add a flag to model with different MSA depth. Switched off by default
* Add AF version to meta.json
* Simplified meta.json structure
* update the urls with https
* Fixed parsing seqs from cif when chaid id is wrong
* Unification: always pass first chain id, then seq
* Oops
* Modify to_mmcif() for TrueMultimer
* Reprt num_msa and num_extra_msa in gradient mode
* Adapt to_mmcif() for use with TrueMultimer
* Point to commit 768f35d
* Different MSA depth for different predictions
* Use a custom class for handling biopython structure
* PEP8
* fix bug wrong longic with author chain
* pytest for create_fake_database.py
* Check number of atoms and that atoms contain only label ids from poly_seq loop
* Debug printing
* Check integrity of _atom_site.label_seq_id in the template mmcifs
* Added test for a bare pdb template
* Remove clashing/low_plddt residues from seqres_to_structure for consistency.
* run_multimer: added model_names and msa_depth flags
* New.
* run_multimer.py: update --gradient_msa_depth description
utils.py: fix unpacking from compute_msa_ranges()
* Update colabfold
* New.
* just renamed fake to custom :-)
* Copy templates to the custom templates to avoid conflicts with files with the same file names
* check that features are generated for mmt using subprocess
* check that template seq is not None, atom positions are not all zeros
* Check non-zero coordinates for backbone atoms for non-gap residues in template
* Parse seqres from pdb template using SeqIO
* Residues with non-zero atom coords are identical for seqatom and seqres
* benchmark models for testing truemultimer
* Only for identical number of atoms for tests
* Create uniprot_runner, otherwise test always fails
* Remove unused imports
* Added test for truemultimer
* Run TM tests on 3L4Q
* test/check_predict_structure.py
* Check that the best RMSD from 3 models is below 3A
* Tests for custom msa depth
* New
* Bumped to v1.00.0
* sync names with the main branch
---------
Co-authored-by: Jan Kosinski <jan.kosinski@embl-hamburg.de>
Co-authored-by: Dmitry Molodenskiy <dmolodenskiy@login01.cluster.embl.de>
Co-authored-by: Dingquan Yu <44894955+dingquanyu@users.noreply.github.com>
