READRetro_data

READRetro_data is for READRetro (https://github.com/SeulLee05/READRetro)All files including data.sh should be located in READRetro/READRetro_data/(here) before running data.sh 

The directory structure of the Zenodo data should be follows after running sh data.sh:

READRetro_data

    ├── data.sh

    ├── data

    │   ├── model_train_data

    │   └── multistep_data

    ├── model

    │   ├── bionavi

    │   ├── g2s

    │   │   └── saved_models

    │   ├── megan

    │   └── retroformer

    │       └── saved_models

    ├── result

    └── scripts

Place the READRetro_data into the READRetro directory (i.e., READRetro/READRetro_data) and run `sh data.sh` in READRetro_data to set up the data.

Ensure the data is correctly located in READRetro. Verify the following:

- `READRetro/retroformer/saved_models` should match `READRetro_data/model/retroformer/saved_models`

- `READRetro/g2s/saved_models` should match `READRetro_data/model/g2s/saved_models`

- `READRetro/data` should match `READRetro_data/data/multistep_data`

- `READRetro/result` should match `READRetro_data/result`

- `READRetro/scripts` should match `READRetro_data/scripts`

The directories `READRetro_data/model/bionavi`, `READRetro_data/model/megan`, and `READRetro_data/model_train_data` are essential for reproducing the manuscript*ckpts.tar.gz has been added. It consists of checkpoints for MHN-React and GraphRetro, which were used in a single-step retrosynthesis evaluation. (v2)

*parameter_optimization.zip has been added. It consists of checkpoints for Retroformer and Graph2SMILES, which were used to construct the ensemble models applied to multi-step retrosynthesis in parameter optimization (model_1600000.pt: Retroformer layout 16, model.72000_35.pt: Graph2SMILES layout 16). (v3)

READRetro_data 是专为 READRetro（https://github.com/SeulLee05/READRetro）设计的数据集。运行 data.sh 前，所有包含 data.sh 的文件均需放置于 READRetro/READRetro_data/ 目录下。

运行 sh data.sh 后，Zenodo 数据集的目录结构应如下所示：

READRetro_data
├── data.sh
├── data
│ ├── model_train_data
│ └── multistep_data
├── model
│ ├── bionavi
│ ├── g2s
│ │ └── saved_models
│ ├── megan
│ └── retroformer
│ └── saved_models
├── result
└── scripts

将 READRetro_data 移入 READRetro 目录（即路径为 READRetro/READRetro_data），并在 READRetro_data 目录中执行 `sh data.sh` 以完成数据集配置。

请确保数据集已正确部署至 READRetro 目录，请核对以下路径匹配情况：
- `READRetro/retroformer/saved_models` 应与 `READRetro_data/model/retroformer/saved_models` 一致
- `READRetro/g2s/saved_models` 应与 `READRetro_data/model/g2s/saved_models` 一致
- `READRetro/data` 应与 `READRetro_data/data/multistep_data` 一致
- `READRetro/result` 应与 `READRetro_data/result` 一致
- `READRetro/scripts` 应与 `READRetro_data/scripts` 一致

`READRetro_data/model/bionavi`、`READRetro_data/model/megan` 及 `READRetro_data/model_train_data` 目录是复现论文实验结果的必备目录。
*ckpts.tar.gz 已新增，其中包含用于单步逆合成（retrosynthesis）评估的 MHN-React 与 GraphRetro 模型检查点。（v2）
*parameter_optimization.zip 已新增，其中包含用于构建多步逆合成（retrosynthesis）集成模型的 Retroformer 与 Graph2SMILES 模型检查点，用于参数优化环节（model_1600000.pt：16层Retroformer架构；model.72000_35.pt：16层Graph2SMILES架构）。（v3）