KKXR

Hi, I have run this command but didn't encounter this problem. However, I noticed that the number of entries is different from which was provided in this repo, and that...

Emmm, this is a bit weird, because if the correct database is provided and --pre_numbered is passed, then the post process shouldn't enter the renumber branch ([line 246 in download.py](https://github.com/THUNLP-MT/MEAN/blob/main/data/download.py#L246))....

I just downloaded the newest version of SAbDab, and the first part of logs is as follows: > PDB file already renumbered with scheme imgt > downloading raw files >...

Oh in that case I know the problem. You need to further download the PDB database annotated by SAbDab ([this link](https://opig.stats.ox.ac.uk/webapps/newsabdab/sabdab/search/?all=true#downloads)). The database is downloaded as compressed file in zip...

The initial error is "ModuleNotFoundError: No module named 'utils'". The folder utils should be at the root directory of this repo. Could you check if it still there in your...

按目前的报错信息我很难确定是什么原因。报错是"ModuleNotFoundError: No module named 'utils'"，但你下载的代码是完整的。同样的情况下我在多台机器上尝试均没有办法复现这个报错。如果你直接在项目根目录下运行python -m data.download是否会报这个错误呢？

Hi, 在[这里](https://github.com/THUNLP-MT/MEAN/blob/main/generate.py#L76)，是生成的sequence和ground truth的sequence的重合比例

我觉得应该可以先借鉴[数据处理的代码](https://github.com/THUNLP-MT/MEAN/blob/main/data/dataset.py#L140)把pdb处理成AAComplex类，然后调用get_cdr()的方法取出CDR的数据，再用get_seq()就可以得到序列，从而计算AAR

这里json load的只是一个summary文件，pdb里存的是具体的结构数据，你应该只需要参考140-147行将pdb转为AAComplex的逻辑就可以了

Hi, sorry for the ambiguity in the instructions. The distribution learning shares the same procedure as in [property prediction](https://github.com/THUNLP-MT/PS-VAE/tree/main/src#training). The dataset and the vocab entry you are using is correct....