[[ https://github.com/uci-cbcl/UFold | UFold Home ]]
<markdown>
# copy the UFold directory to your workspace
UFold is installed on white.

```
cp -r /home/usr/share/UFold ~/UFold
cd ~/UFold
ls 
# common  data  models  readme.md  results  run.sh  ufold  ufold_predict.py"
```
You should not change these files:
- common + ufold: files about the architecture of UFold
- models: pretrained weight files
- ufold_predict.py: the real script to predict the secondary structure

You can check these files:
- data: directory to put your sequence in
- results: directory that contains the prediction results
- run.sh: the direct script to be implemented
- readme.md: some explanations


# put your fasta file into the directory 'data' and rename it to input.txt
```
cp YOURFILE ./data
unlink input.txt # you may need to delete the previous one
ln -s data/YOURFILE data/input.txt
```

# run the script
```
./run.sh
# You shall modify it for yourself

# You may encounter the warning 
# '/home/xuelab/softwares/miniconda3/envs/UFold/lib/python3.6/site-packages/torch/nn/functional.py:718: UserWarning: Named tensors......'
# just ignore it.

# Your results are in the directory 'results', with the same name every time...
# So keep it carefully, every time you run a new job.
```
</markdown>