Duplicate reaction families? Surface_Dual_Adsorption_vdW and Surface_Abstraction_Beta_double_vdW

[rwest]

I think that this reaction matches the family Surface_Dual_Adsorption_vdW
[<Entry index=23 label="C=C">, <Entry index=55 label="C-R">]

where R4-R6 are a C-H (note that I am using the proposed corrected chemdraw picture, see #630 )

but its reverse reaction matches Surface_Abstraction_Beta_double_vdW
[<Entry index=3 label="C-H">, <Entry index=2 label="Adsorbate1">]

The Surface_Dual_Adsorption_vdW family was added in #385 (specifically commit 8f63ed8)

The Surface_Abstraction_Beta_double_vdW family was added in #521

Both additions seem to use this example reaction: COOH* + OH* <=> CO2* + H2O*

We should resolve this duplication (consolidate any useful data and remove one family)
and should implement database tests to detect and prevent other inadvertent duplication.

[bjkreitz]

I created the Surface_Abstraction_Beta_double_vdW family because we have BEP relations from the literature for this direction. The Surface_Dual_Adsorption_vdW family is just based on one training reaction. I trust the results from the Surface_Abstraction_Beta_double_vdW more and never used Surface_Dual_Adsorption_vdW. Therefore, I would suggest to remove Surface_Dual_Adsorption_vdW.

[rwest]

Thanks Bjarne.
The plot thickens...
Some reactions are matching both Surface_Abstraction_Beta_double_vdW AND Surface_Dual_Adsorption_vdW like

But some reactions are matching only Surface_Dual_Adsorption_vdW like

Can these reactions occur? (they do show up in Magraf's list). If so, should/can we modify Surface_Abstraction_Beta_double_vdW so that it matches/generates them?

[ChrisBNEU]

for the first reaction, the "abstracting" is R!H so it's disallowed. H is technically abstracting the methyl group.

1 *6 Xo  u0 {2,S}
2 *4 R!H u0 px cx {1,S}	

Second, third, and fourth, in the "donating" group, we have a double bond between *1 and *2 in the donating group. we don't allow a double bond to move to a triple.

1 *1 R!H u0 px cx {2,S} {4,S}
2 *2 R!H u0 px cx {1,S} {3,S}
3 *3 R   u0 {2,S}
4 *5 Xo  u0 {1,S}

In general, maybe we could make this family more permissive? I am not sure if that is opening us up for more junk kinetics but we should probably err on the side of allowing more.

[bjkreitz]

I think we can make the Surface_Abstraction_Beta_double_vdW family more permissive. However, an H atom shouldn't be able to abstract a CH3 group, but the other reactions are reasonable (although very funky). I think we should allow double bonds between 1 and 2. I'm going to update the PR.