Open kimdn opened 1 week ago
Can you please clarify "and then copy the source codes of HighFold into the installed ColabFold preject."
?
Even if I remove HighFold folder from my localcolabfold folder,
input_fasta="/people/kimd999/script/python/predict_structure/protein_only/HighFold/input.fasta"
colabfold_batch --model-type alphafold2 $input_fasta path_output
still runs.
However, I cannot reproduce your paper's HighFold results with and
I could reproduce your
and
HighFold results only.
I installed localcolabfold and copied HighFold git repo into my localcolabfold.
So that my localcolabfold git repo looks as
Then I ran
input_fasta="/people/kimd999/script/python/predict_structure/protein_only/HighFold/input.fasta" colabfold_batch --model-type alphafold2 $input_fasta path_output
where my input.fasta looks
>fasta XXXXXXXXXX
(here 'X' means any AA of mine. My input sequence has only 1 cysteine. Therefore, I don't expect internal S-S bond)
However, all 5 predicted peptide structures are linear (i.e., not expected cyclic peptide).
I wonder whether I run HighFold correctly. Or I ran HighFold well, just my input AA sequence is indeed predicted to be linear according to HighFold.
Can you please clarify
"and then copy the source codes of HighFold into the installed ColabFold preject."
?Even if I remove HighFold folder from my localcolabfold folder,
input_fasta="/people/kimd999/script/python/predict_structure/protein_only/HighFold/input.fasta" colabfold_batch --model-type alphafold2 $input_fasta path_output
still runs.
However, I cannot reproduce your paper's HighFold results with and
I could reproduce your
and
HighFold results only.
You should replace the colabfold and alphafold files provided by the author into the localfold directory. The specific location is: /your/path/localcolabfold/colabfold-conda/lib/python3.xx/site-packages/colabfold & alphafold. Remember to backup the files that will be replaced before making the changes. After the replacement, some package imports might show errors. You can modify these based on the original files.
I installed localcolabfold and copied HighFold git repo into my localcolabfold.
So that my localcolabfold git repo looks as
Then I ran
where my input.fasta looks
(here 'X' means any AA of mine. My input sequence has only 1 cysteine. Therefore, I don't expect internal S-S bond)
However, all 5 predicted peptide structures are linear (i.e., not expected cyclic peptide).
I wonder whether I run HighFold correctly. Or I ran HighFold well, just my input AA sequence is indeed predicted to be linear according to HighFold.