# Data
## Structures
The matrices for 3 target structures (to synthesize) and a database of 7165 query structures (to combine to build the target)
are compressed in `data.npz` 

Within python, it can be read like: 
```
data = np.load("data.npz", allow_pickle=True)
```

where `data.files` will return the names of the numpy arrays (should be `target_labels, target_CMs, target_ncharges, database_labels, database_CMs, database_ncharges`) 
where CMs are the matrices (of target and database respectively) and the corresponding arrays can be accessed like: 

```
data["target_labels"]
```

For more details see the documentation: 
https://het.as.utexas.edu/HET/Software/Numpy/reference/generated/numpy.savez.html

## Connectivity / functional group information
Adjacency matrices and functional group information derived from the connectivity are compressed in `connectivity_data.npz`. 

Within python, it can be read like:
```
connectivity_data = np.load("connectivity_data.npz")
```

the corresponding keys are `fg_counts_targets` for the functional group counts of each of the 3 target molecules,`fg_counts_frags` for the functional group counts of
each of the fragment molecules, `frag_adj_matrices` for the adjacency matrices of the fragments and `target_adj_matrices` for the adjacency matrices of the target molecules.
The order is the same as those in `data` containing the structures.