This readme contains all of the information that is used for the manuscript. 
Information about each folder is given below, along with folder structure.

---building_blocks.zip
   ---building_blocks
       |--- This folder contains all of the XYZs (appended with the nonperiodic molecular graph) for building blocks used in this work.
   ---building_block_aliases
       |--- This folder contains three CSV files (inorganic, organic, and edge). Each CSV file contains the individual building blocks for each category, and important information for the building blocks. For instance, both node CSV files contain each node, the nets that it is testable by (purely considering connection points and connectivity --> denoted as testable nets), the number of nets the node is actually compatible with, and its corresponding alias and molecular graph determinant. The "X_count" refers to the number of connection points for a given node. This folder contains the mappings between building blocks and the MOFs they came from.

---features_and_properties.zip
   ---features_and_predictions
       |--- This folder contains all features and predictions of ML models.
       |------BWDB --> features and predictions for Boyd and Woo DB of 300K structures.
       |------hMOF --> features and predictions for Wilmer DB of 100K structures.
       |------ToBaCCo --> features and predictions for Colon and Gomez-Gualdron DB of 12K structures.
       |------ultrastable --> features and predictions for 50K MOFs constructed from ultrastable parts, split into the type of MOF (1 inor+1 edge, 2 inor+1 edge, 1 inor+1 org+ 1edge)
       |------ARC --> features and predictions for the MOFs in the ARC MOF database.
    ---computed_properties
       |--- Computed properties on the set of MOFs deemed to be ultrastable. Additionally, a set of ~1000 thermally and activation unstable MOFs obtained from k-medoids sampling is provided for analysis. For methane adsorption, we provide uptake values, which can be converted into working capacities by taking the difference of high and low pressure uptakes.


---initial_structures.zip
   ---initial_structures
       |--- Building block constructed CIF files prior to any force field optimization. 

---optimized_structures.zip
   ---optimized_structures
       |--- All structures that are successfully optimized using the procedure described in the main text. These structures are attempted for featurization.

Note: An edge of "none" means that the MOF does not have any edges at all. This means that a subset of the structures in 1 inor-1 edge are actually 1 inorganic node and 0 edges. Same holds true for 1 inor-1 org-1 edge. These MOFs can be identified readily by looking for edges of none. ANY MOF can be constructed with an edge of "none".