redundant link in martini3001

[fgrunewald]

This link is never applied because it get's overwritten by another link. This is the intended behavior and just a remainder of development.

vermouth-martinize/vermouth/data/force_fields/martini3001/aminoacids.ff

Lines 1091 to 1099 in b450b67

	[ link ]
	[ angles ]
	; (Modified by Paulo)
	-BB BB +BB 10 127 25 {"group": "BBB angles"}
	[ patterns ]
	-BB BB {"cgsecstruct": "C", "resname": "PRO|HYP"} +BB
	-BB {"cgsecstruct": "C", "resname": "PRO|HYP"} BB +BB
	-BB BB +BB {"cgsecstruct": "C", "resname": "PRO|HYP"}

[pckroon]

I don't see which one overwrites it?

[fgrunewald]

yep really hard to see. Took me like 1h to figure it out. But this one:

vermouth-martinize/vermouth/data/force_fields/martini3001/aminoacids.ff

Lines 1128 to 1137 in b450b67

	[ link ]
	resname $protein_resnames
	[ angles ]
	; (Modified by Paulo)
	-BB BB +BB 10 127 20 {"group": "BBB angles"}
	[ patterns ]
	-BB BB {"cgsecstruct": "C", "resname": $protein_resnames_non_pro} +BB
	-BB {"cgsecstruct": "C", "resname": $protein_resnames_non_pro} BB +BB
	-BB BB +BB {"cgsecstruct": "C", "resname": $protein_resnames_non_pro}
	-BB {"cgsecstruct": null} BB {"cgsecstruct": null} +BB {"cgsecstruct": null}

The reason is that this link comes after the previous one

[pckroon]

But the patterns are different. The first applies to coil PRO/HYP, the second to coil non-PRO.

[fgrunewald]

yes but the bottom patterns include all above patterns; the order in the ff file is incorrect

[pckroon]

I don't think it does?
-BB BB {"cgsecstruct": "C", "resname": "PRO|HYP"} +BB <!=
-BB BB {"cgsecstruct": "C", "resname": $protein_resnames_non_pro} +BB
Note that it says protein_resnames_non_pro, so anything but PRO and HYP.

[fgrunewald]

Let's say we have the sequence ALA-PRO-ALA. What we expect is to have an angle 127 25 due to the first link. That matches because we have a Proline in the middle. But this match gets overwritten by the second pattern of the other link (i.e. -BB {"cgsecstruct": "C", "resname": $protein_resnames_non_pro} BB +BB ) This pattern only requires the first BB bead to be non-proline, which is true. Thus we get 127 20. This is true for ALL cases. Because we never noticed, all itp files ever produced did not apply the first link, which therefore makes it redundant.

[pckroon]

Ooooooh you're right. Good catch :)

[fgrunewald]

@pckroon do we remove it or leave it?

[pckroon]

Remove it, since it slows down the FF