About bbs and topo

[Yong-Q]

I would like to know how the bbs and topo in the COF database(or other) are decomposed. Do you decompose by cif?

Thanks!

[Yeonghun1675]

Hi @Yong-Q,

Can you elaborate on what you mean by BBS and TOPO being disassembled, do you mean when you create the BBS and TOPO database?

[Yong-Q]

Right, I just wanted to ask you about how to create BBS ahd TOPO database.

Best wishes!

[Yeonghun1675]

For the topology, we used the cgd file available in the RSCR database (https://rcsr.anu.edu.au/). For 2D topology, we converted it to 3D and used it. For building blocks, we used 130 genetic structures provided by Lan et al ( Materials genomics methods for high-throughput construction of COFs and targeted synthesis), we used the 130 genetic structure units provided by Lan et al, adding functional groups.

We did not decompose the building blocks separately. If you want to decompose the building blocks, you can use the decomposer provided by pormake. (https://github.com/Sangwon91/PORMAKE/blob/main/pormake/experimental/decomposer/README.md)

[Yong-Q]

For Topo, it is not the case that every porous material is consistent, whether COF or MOF

[Yeonghun1675]

What exactly do you mean by "not consistent" - do you mean different types of topologies used, or do you mean that the same topology can be used differently?

For topoolgy types, the type used for each material in the synthesized database can be different. In our work, we dealt with hypothetical porous materials, so we assumed that we could create hypothetical topologies that were not synthesized. Therefore, we created hypothetical porous materials regardless of the actual topology used. This is because the porous materials we created are only used in the pre-training process for PMTransformer, so we do not need strict criteria for each material. If each material needs to be created with strict criteria, such as a generative model, the topology can be determined using a program such as topospro after decomposition into building blocks.

The different use of topology seems to depend on the building block. If the building block is a complex conformation, it may be determined to be a different topology than the one previously generated by pormake according to the strict definition of topology. (see: figure 1 of MOFId paper https://pubs.acs.org/doi/10.1021/acs.cgd.9b01050) In our work, the topology was determined as single node, and the topology may be different if used with other methodologies (e.g. all node, metal-oxo). However, we believe that this depends on the basis of dividing the building blocks and not on the porous material.

[Yong-Q]

I see what you mean. The key to distinguish MOF from COF is whether there are metal nodes in bbs. Thanks!