As a validation of the method used to construct the database of
zeolite-like materials, we note that the computational search through
crystallographic space successfully found 83% of the known zeolite
topologies. Several topologies that were not found have a greater number
of crystallographically unique atoms than the maximum number that was used
in the search and so are automatically not present in the current
database. We performed a comparison of the unit cell parameters for known
zeolites among the 1) the International Zeolite Association (IZA)
database, 2) the zeolites found in the zeolite-like database, and 3)
zeolites minimized by GULP starting from the IZA structure.  In the latter
two cases, we compared with both the SLC and BKS interatomic potentials. 
The results of this comparison can be found in this supplementary
information. The computational search method works remarkably well,
matching the unit cells of the IZA structures within +/- 1% of the a, b,
c, \alpha, \beta, and \gamma values in most instances, suggesting that a
crystallographic search of the database for a newly found material may be
an effective route to structure determination.

The program PLATON was used to find primitive unit cells. We note that in
some cases, the known zeolites were found in the database in lower symmetry
settings (e.g. SOD). In other cases. known zeolites were found with
supercells in the database, and these supercells were reduced to the
primitive unit cell whenever possible.  If results from PLATON were
modified, the results from PLATON are listed in the _orig.dat file.   
