Prop3D Dataset Organization

CATH-based Dataset

The inherently hierarchical structure of CATH (A) is echoed in the design schema underlying the Prop3D dataset (B), as illustrated above. Prop3D can be accessed through a cloud-based HDF file (HSDS) seeded with the CATH hierarchy for all available superfamilies. For clarity, an example of one such superfamily is the individual H-group 2.60.40.10 (Immunoglobulins) shown here as the orange sector (denoted by an asterisk near 4 o’clock). Each such superfamily is further split into:

1. The domain groups, with datasets provided for each domain (atomic features, residue features, and edge features), as delineated in the upper-half of (B)

2. Pre-calculated data splits (data_splits), shown in the lower-half of (B), which exist as hard-links (denoted as dashed green lines) to domain groups.

3. Pre-calculated representatives for each superfamily, provided by the CATH, and are hard-linked to domain groups.

Each domains is further separated into:

1. atoms, a dataset for atom-based coordinates and features

2. residues, a dataset for residue-based coordinates and features (useful for coarse-gaining or graph based applications)

3. edges, a dataset for residue-residue interaction graphs with edge features

The ‘sunburst’ style CATH diagram, from http://cathdb.info, is under the Creative Commons Attribution 4.0 International License.

Prop3D-20 Dataset

For the initial release of Prop3D, we include Prop3D-20, a condensed dataset of just 20 superfamilies with high-population density or of interest to our DeepUrfold project. The HSDS H5 is organized as follow:

We organize each superfamily by their CATH hierarchy within the HDF file:

/1
    /10
        /10
            /10
                /domains
                /data_splits
                /representatives
        /238
            /10
                /domains
                /data_splits
                /representatives
        /490
            /10
                /domains
                /data_splits
                /representatives
        /510
            /10
                /domains
                /data_splits
                /representatives
    /20
        /1260
            /10
                /domains
                /data_splits
                /representatives
/2
    /30
        /30
            /100
                /domains
                /data_splits
                /representatives
    /40
        /50
            /140
                /domains
                /data_splits
                /representatives
    /60
        /40
            /10
                /domains
                /data_splits
                /representatives
/3
    /10
        /20
            /30
                /domains
                /data_splits
                /representatives
    /30
        /230
            /10
                /domains
                /data_splits
                /representatives
        /300
            /20
                /domains
                /data_splits
                /representatives
        /310
            /60
                /domains
                /data_splits
                /representatives
        /1360
            /40
                /domains
                /data_splits
                /representatives
        /1370
            /10
                /domains
                /data_splits
                /representatives
        /1380
            /10
                /domains
                /data_splits
                /representatives
    /40
        /50
            /300
                /domains
                /data_splits
                /representatives
            /720
                /domains
                /data_splits
                /representatives
    /80
        /10
            /10
                /domains
                /data_splits
                /representatives
    /90
        /79
            /10
                /domains
                /data_splits
                /representatives
        /420
            /10
                /domains
                /data_splits
                /representatives

PDB-based Dataset

Still in progress.

The PDB-based datasets is organized like a single superfamily just domains, data_splits and representatives at the root level. Representatives are taken from the first entry of each mmseqs 30% clusters provided by the PDB.