dssr parameters

[auffinger]

Hi Xiang-Jun,

The X3DNA package came with a configuration file named 'misc_3DNA.par'.
You showed me once how to precise the distance limit for hydrogen bonds in dssr.
What about the other parameters like those related to
the vertical base separation,
the maximum distance between base origin,
the maximum vertical separation,
...

Hope equivalent parameter sets are available in dssr.

Best regards,

Pascal

[xiangjun]

Yes, 'misc_3dna.par' settings are mostly available in DSSR through command line. I've deliberately not documented them (plus quite a few new ones) in the current beta releases to ensure the defaults are working well for the majority of cases, and to make DSSR less overwhelming/intimidating to first time users. DSSR has more to offer than meets the eye.

Here are the ones you asked:

--pair_vd2   the vertical base separation
--pair_od2   the maximum distance between base origin

I do not understand what you mean by the following:

the maximum vertical separation

HTH,

Xiang-Jun

[auffinger]

Thanks Xiang-Jun,

In X3DNA, you had besides the ones you provided, the followings.

Parameters for defining bifurcated h-bonds,
The possibility to define atoms involved in h-bonding (O,N, or C)

The maximum vertical base separation (sorry for not having written the entire line down)
The maximum angle between base normals
The minimum distance between the RN/YN base atoms

Thanks for sharing more if you like,

Best,

Pascal

[auffinger]

Sorry,

forgot the "#   maximum angle between base normals (in range 0..90)"

Pascal

[xiangjun]

Parameters for defining bifurcated h-bonds,

No longer needed in DSSR which always finds bifurcated h-bonds.

The possibility to define atoms involved in h-bonding (O,N, or C)

Check the --chbond option (experimental)

The maximum vertical base separation (sorry for not having written the entire line down)

Already answered: --pair_vd2

The maximum angle between base normals
…
forgot the "#   maximum angle between base normals (in range 0..90)"

Check the --pair_angle2 option

The minimum distance between the RN/YN base atoms

No longer used.

HTH,

Xiang-Jun

[auffinger]

Great Xiang-Jun,

Is there a possibility to set up different cut-offs for O/N...H and C-H...N bonds ?
That would be really awesome.

Best,

Pascal

[xiangjun]

No need to set cut-off for C-H…N/O bond: DSSR uses a more stringent criteria for such H-bonds internally.

As a rule, please provide concrete examples: "That would be really awesome" than just making a general request. Where do you need a separate set of cut-offs for C-H containing H-bonds?

Xiang-Jun

[auffinger]

Hi Xiang_Jun,

Thanks again for the help.
A last request would be to know what the default parameters are for these options (may be they could be
written down along with the current values when the "note" option is activated.

As for H-bonds, I used longer cut-offs for C-H...O bonds than for regular ones.
Usually for base pairs, I use the 3.35 and for C-H...O bonds, I add 0.5 Å to the preceding value.

Thus, a different cut-off for C-H...O bonds could be valuable at some point.
Similarly, a specific cut-off for other atoms like S could be interesting.

Best,

Pascal

[xiangjun]

Then you'd also use different distance cutoffs for Halogen bonding involving F, Cl, Br, and I. However, DSSR is first and foremost for deriving the secondary structure of RNA from three-dimensional coordinates. As we discussed before, 3DNA/DSSR is not a dedicated software for finding various H-bonds. Have you ever considered HBPLUS (http://www.ebi.ac.uk/thornton-srv/software/HBPLUS/), among other possible choices?

Xiang-Jun

[auffinger]

OK, thanks, what about default values ?

Pascal

[xiangjun]

Regarding the default values of the parameters, they are the same as in 3DNA 'misc_3dna.par' -- 4.0 Å for H-bond upper limit, 2.5 Å for vertical base separation, 15.0 Å for maximum distance between base origins, and 65° for maximum angle between base normals.

As for H-bonds, I used longer cut-offs for C-H...O bonds than for regular ones.
Usually for base pairs, I use the 3.35 and for C-H…O bonds, I add 0.5 Å to the preceding value.

DSSR uses a different strategy for C-H…O/N bonds and the criteria are more stringent than for the canonical H-bond. Note also that DSSR also identifies C-H…N bonds, just as for C-H…O bonds. I am more conservative in such non-canonical cases -- I was once even lectured about the existence of the O2′(G)…O2P(U) H-bond in the GpU dinucleotide platforms. Instead of arguing at the methodical level, I'd ask you to provide concrete examples where DSSR misses obvious C-H…O/N bonds.

While I am open to suggestions to make 3DNA/DSSR more generally applicable, I am very careful in selecting what to be put into the software that I support. To me, it is better to have extra easter eggs than claiming more than a piece of software can actually do. After all, only you know exactly what you are looking for. Understandably, DSSR does not necessarily fits all your requirements. Just use the parts you find useful, look elsewhere or develop your own tools to get the job done.

HTH

Xiang-Jun