mmcif_cache

[vetmax7]

Hello!

To train FastFold we need mmcif_cache. How can I generate it? In train.py:

parser.add_argument( "--template_release_dates_cache_path", type=str, default=None, help="""Output of scripts/generate_mmcif_cache.py run on template mmCIF files."""

But in scripts, there is no generate_mmcif_cache.py

Also in train.sh:

mmcif_cache.json
chain_data_cache.json

[morou34]

@vetmax7 Hi ; I was wondering if you solved this issue as i'm currently facing the same thing; thanks;

[vetmax7]

@vetmax7 Hi ; I was wondering if you solved this issue as i'm currently facing the same thing; thanks;

Hello!

I used scripts from OpenFold

[morou34]

@vetmax7 thank you, i found that solution the day i asked you, but thanks for your reply.

[morou34]

@vetmax7 Hi again I was wondering if you managed to launch the training script; i need some help if you were able to do it.

[vetmax7]

@morou34 Hello! I've not run training script of FastFold because it requires internet connection during this but my Univ Cluster does not provide it. I use OpenFold instead.

[morou34]

@vetmax7 Thank you for your answer; do you know why does it require internet exactly (for what) ?

[vetmax7]

As I understood, colossalai libs require internet connection.

[morou34]

I was wondering if i can get your contact, an email or something to contact you a little bit faster; thank you so much for your help and time