training model

[fangyingdawei]

It's a great honor to read your article, but I really hope to see your training network. I would be grateful if you could provide it！

[fusong-ju]

Sorry to say that it is not easy to provide feasible training code due to some license limitations. I can give you some sources.

1. Dataset. CATH data could be downloaded via https://www.cathdb.info and protein structures could be downloaded via https://www.rcsb.org.
2. Input features. The only input features are produced by DeepMSA( https://zhanglab.dcmb.med.umich.edu/DeepMSA/).
3. Training code. You can follow the Method section in the paper (network structure code could be dumped from the checkpoint file).

Feel free to ask me if you need any details.

BTW, I suggest using AlphaFold for better performance

On Sun, Feb 19, 2023 at 6:41 PM fangyingdawei @.***> wrote:

It's a great honor to read your article, but I really hope to see your training network. I would be grateful if you could provide it！

— Reply to this email directly, view it on GitHub https://github.com/fusong-ju/ProFOLD/issues/5, or unsubscribe https://github.com/notifications/unsubscribe-auth/AQBVMFFOOF4QVYVG6N5NLULWYH2FNANCNFSM6AAAAAAVA4MWIA . You are receiving this because you are subscribed to this thread.Message ID: @.***>

[fangyingdawei]

非常感谢您的回复！

------------------ 原始邮件 ------------------ 发件人: "fusong-ju/ProFOLD" @.>; 发送时间: 2023年2月21日(星期二) 中午11:45 @.>; @.**@.>; 主题: Re: [fusong-ju/ProFOLD] training model (Issue #5)

Sorry to say that it is not easy to provide feasible training code due to some license limitations. I can give you some sources.

1. Dataset. CATH data could be downloaded via https://www.cathdb.info and protein structures could be downloaded via https://www.rcsb.org.
2. Input features. The only input features are produced by DeepMSA( https://zhanglab.dcmb.med.umich.edu/DeepMSA/).

3. Training code. You can follow the Method section in the paper (network structure code could be dumped from the checkpoint file).

   Feel free to ask me if you need any details.

   BTW, I suggest using AlphaFold for better performance

   On Sun, Feb 19, 2023 at 6:41 PM fangyingdawei @.***> wrote:

   > It's a great honor to read your article, but I really hope to see your > training network. I would be grateful if you could provide it！ > > — > Reply to this email directly, view it on GitHub > <https://github.com/fusong-ju/ProFOLD/issues/5&gt;, or unsubscribe > <https://github.com/notifications/unsubscribe-auth/AQBVMFFOOF4QVYVG6N5NLULWYH2FNANCNFSM6AAAAAAVA4MWIA&gt; > . > You are receiving this because you are subscribed to this thread.Message > ID: @.***> >

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>

[kkz715]

很抱歉，由于 一些许可证限制。我可以给你一些消息来源。 1. 数据集。CATH 数据可以通过 https://www.cathdb.info 和 蛋白质结构可以通过 https://www.rcsb.org 下载。 2. 输入功能。唯一的输入特征由DeepMSA（https://zhanglab.dcmb.med.umich.edu/DeepMSA/）产生。 3.训练代码。您可以按照论文中的方法部分（网络 可以从检查点文件转储结构代码）。 如果您需要任何详细信息，请随时询问我。 顺便说一句，我建议使用AlphaFold以获得更好的性能 ... On Sun, Feb 19, 2023 at 6:41 PM fangyingdawei @.> wrote: It's a great honor to read your article, but I really hope to see your training network. I would be grateful if you could provide it！ — Reply to this email directly, view it on GitHub <#5>, or unsubscribe https://github.com/notifications/unsubscribe-auth/AQBVMFFOOF4QVYVG6N5NLULWYH2FNANCNFSM6AAAAAAVA4MWIA . You are receiving this because you are subscribed to this thread.Message ID: @.>

Hello author, may I ask you about training structure code.How to get the network structure code from the checkpoint file?