Table 4 Most important 1D/2D birth–death points for the different datasets (in Angstroms). These values correspond to the porous framework sizes most important for a given adsorption task.

From: Machine learning with persistent homology and chemical word embeddings improves prediction accuracy and interpretability in metal-organic frameworks

Target property

1D birth

1D death

2D birth

2D death

(a) BW dataset

log(K\(_{H}\)) CO\(_{2}\)

1

4

3.3

4.1

log(K\(_{H}\)) CH\(_{4}\)

1.6

2

3.6

4.4

0.15 bar CO\(_{2}\)

3.5

3.6

3.4

4

16 bar CO\(_{2}\)

1.7

2

3.1

3.9

5.8 bar CH\(_{4}\)

1.4

3

3.8

4.6

65 bar CH\(_{4}\)

3.6

4.3

2.3

3.2

(b) CoREMOF dataset

log(K\(_{H}\)) CO\(_{2}\)

0.3

1.3

2.3

3.1

log(K\(_{H}\)) CH\(_{4}\)

0.3

1

3.6

4.4

5.8 bar CH\(_{4}\)

1

3.3

3.4

4

65 bar CH\(_{4}\)

3.9

4.8

2.4

3.2

(c) hMOF dataset

0.01 bar CO\(_{2}\)

0.02

0.7

3.2

3.5

0.05 bar CO\(_{2}\)

1.1

1.6

1.6

2.1

0.1 bar CO\(_{2}\)

1.1

2.7

4.4

5.5

0.5 bar CO\(_{2}\)

1.3

3.5

4.7

5.8

2.5 bar CO\(_{2}\)

1

3.7

4

5.1