Delineating the components of bulges, internal loops, junctions, etc

[jyvdf3asdg2]

Hi again Xiang-Jun,

Hopefully you are not tired of my comments and suggestions by now!

I was recently looking through a couple RNA structures looking for a bulge with a non-paired adenosine
and found that it was quite difficult to discern what is what in bulges/internal loops/junctions
just by reading the list of residues. Meaning, which residues are on one strand or the complement, or which residue in a [1x0] bulge
are not paired, etc.

To give an example, a simple bulge from 1S72:

List of 12 bulge(s)
   1 bulge: 6 nts; [2x0]; linked by [#19, #-8]
       0.C245+0.G246+0.A247+0.A248+0.U265+0.G266 [CGAAUG]
       
       
As the residues are listed all on one line, it is not immediately apparent which residues are on which
strand and which 2 residues are "flipped out" without going in to the PDB and analyzing the structure.

Would there be any way to have the program show this distinction?

IE:
   1 bulge: 6 nts; [2x0]; linked by [#19, #-8]
       2 bulge bases [GA]; 0.C245+(0.G246+0.A247)+0.A248 [C(GA)A]
       0 bulge bases; 0.U265+0.G266 [UG]

The same concept holds for internal loops and junctions, basically any structural element that uses the [AxB(xC)] nomenclature. It would, to a layman, make it
much simpler to analyze if the component strands and unpaired/mispaired residues were delineated in some way.

Perhaps I'm just not that well versed in RNA secondary structure vocabulary and the like, but I found it somewhat hard to discern this distinction without a bit more digging.

Thanks so much.

[xiangjun]

Thanks for your feedback. As always, the more, the merrier! DSSR is currently in beta, and I am fully open to user suggestions for every aspect of the software, including format changes.

In writing DSSR, I came across many basic concepts in the literature of RNA structures that are well-known, yet not clearly defined (at least to my understanding). I tried to follow the conventions (where exist) as much as practical, and came up with my own 'ways' where necessary. It is my hope that DSSR would help in establishing pragmatic ways in charactering RNA (secondary) structures.

For your example case:

Code: [Select]

List of 12 bulge(s)
   1 bulge: 6 nts; [2x0]; linked by [#19, #-8]
       0.C245+0.G246+0.A247+0.A248+0.U265+0.G266 [CGAAUG]
It means that DSSR detects 12 bulges in the structure (1s72). For each bulge, DSSR outputs its type as a special case of internal loops with one strand containing 0 nts. Here the first bulge contains 2 nts, so of type [2x0]. Plus two enclosing canonical bps (here from stem #19, and lone bp #-8), the loop contains 6 nts (2+2+2). The listing below contains the 6 nts in sequential order, as would be obvious if one extracts just the nts and displayed them in Jmol or PyMOL. Here the two bulged out nts are 0.G246+0.A247.

I see your point for making components of the loop explicit, and will consider to implement extra output fields in future release of DSSR.

Xiang-Jun

[jyvdf3asdg2]

Thanks for your feedback. As always, the more, the merrier! DSSR is currently in beta, and I am fully open to user suggestions for every aspect of the software, including format changes.

In writing DSSR, I came across many basic concepts in the literature of RNA structures that are well-known, yet not clearly defined (at least to my understanding). I tried to follow the conventions (where exist) as much as practical, and came up with my own 'ways' where necessary. It is my hope that DSSR would help in establishing pragmatic ways in charactering RNA (secondary) structures.

For your example case:

Code: [Select]

List of 12 bulge(s)
   1 bulge: 6 nts; [2x0]; linked by [#19, #-8]
       0.C245+0.G246+0.A247+0.A248+0.U265+0.G266 [CGAAUG]
It means that DSSR detects 12 bulges in the structure (1s72). For each bulge, DSSR outputs its type as a special case of internal loops with one strand containing 0 nts. Here the first bulge contains 2 nts, so of type [2x0]. Plus two enclosing canonical bps (here from stem #19, and lone bp #-8), the loop contains 6 nts (2+2+2). The listing below contains the 6 nts in sequential order, as would be obvious if one extracts just the nts and displayed them in Jmol or PyMOL. Here the two bulged out nts are 0.G246+0.A247.

I see your point for making components of the loop explicit, and will consider to implement extra output fields in future release of DSSR.

Xiang-Jun

Hi Xiang-Jun,

Thanks for the reply and clarification.

It makes sense, but when looking searching for a specific case (IE. a "flipped-out- adenosine in a bulge) it is not as straight-forward as it could be as one has to make inferences.
When talking to some RNA people in my lab about it and showing them the output, they were a bit confused as well at first as it was not explicitly delineated.

If I am quickly looking through a file for a 2-bulge, your program makes it imminently easy to find that. I just look for [2x0].

However, if one wants to go a step further and find a [2x0] containing a non-paired adenosine, it becomes a bit more difficult when reading a concatenated line like: 0.C245+0.G246+0.A247+0.A248+0.U265+0.G266

Thanks again

[xiangjun]

Okay, how about junction loops, and stem loops? Do you have any suggestions for formatting? I'd like to consider all loops in a consistent frame work.

Up to this point, I've been focusing on the underlying algorithm for identifying the various RNA structural components, and their interactions. I'd consider carefully user requests to the extent that's makes sense to me in a global/general sense. Bear in mind, though, some purpose specific scripting/programming may well be necessary.

Xiang-Jun

[jyvdf3asdg2]

Hi Xiang-Jun,

Let me ask around the lab and get their input and I'll make a suggestion based on what they suggest that would be most "user-friendly" and easily readable.

For internal loops, the same suggestion above would be nice, IE. for 1S72:

  55 asymmetric internal loop: 7 nts; [1x2]; linked by [#168, #169]
       0.G2855+0.A2856+0.C2857+0.G2900+0.C2901+0.A2902+0.C2903 [GACGCAC]

Would be readable as:

  55 asymmetric internal loop: 7 nts; [1x2]; linked by [#168, #169]
       1 loop bases [A]; 0.G2855+(0.A2856)+0.C2857 [G(A)C]
       2 loop bases [CA]; 0.G2900+(0.C2901+0.A2902)+0.C2903 [G(CA)C]

Thanks

[xiangjun]

Let me ask around the lab and get their input and I'll make a suggestion based on what they suggest that would be most "user-friendly" and easily readable.

That's great -- once I get your feedbacks, I will implement them in the next beta release of DSSR. I'm planning to add some new features to DSSR, including detection of kink-turns, but would take priority on refinements of existing functionality.

  55 asymmetric internal loop: 7 nts; [1x2]; linked by [#168, #169]
       1 loop bases [A]; 0.G2855+(0.A2856)+0.C2857 [G(A)C]
       2 loop bases [CA]; 0.G2900+(0.C2901+0.A2902)+0.C2903 [G(CA)C]

I like your idea for internal loop. Maybe stem loop, bulge, and junctions can follow the same pattern?

Xiang-Jun

[jyvdf3asdg2]

Let me ask around the lab and get their input and I'll make a suggestion based on what they suggest that would be most "user-friendly" and easily readable.

That's great -- once I get your feedbacks, I will implement them in the next beta release of DSSR. I'm planning to add some new features to DSSR, including detection of kink-turns, but would take priority on refinements of existing functionality.

  55 asymmetric internal loop: 7 nts; [1x2]; linked by [#168, #169]
       1 loop bases [A]; 0.G2855+(0.A2856)+0.C2857 [G(A)C]
       2 loop bases [CA]; 0.G2900+(0.C2901+0.A2902)+0.C2903 [G(CA)C]

I like your idea for internal loop. Maybe stem loop, bulge, and junctions can follow the same pattern?

Xiang-Jun

Hi Xiang-Jun,

Thanks first for considering user input in to how the output is formatted.

I talked to a few involved with RNA, and the first thing they agreed upon was that the bulges, internal loops, junctions and (to a lesser extent) ribose zipper would benefit from being more explicit in which residues belonged to which strand or junction. More specifically, each strand/junction is on a separate line.

According to the input above, this slightly revised example for seems to be easy to understand for junctions/loops/bulges.

Code: [Select]

  55 asymmetric internal loop: 7 nts; [1x2]; linked by [#168, #169]
       1 loop bases 0.A2856 [A]; 0.G2855+0.A2856+0.C2857 [GAC]
       2 loop bases 0.C2901+0.A2902 [CA]; 0.G2900+0.C2901+0.A2902+0.C2903 [GCAC]

  20 3-way junctions: 12 nts; [1x1x4]; linked by [#138, #-43, #140]
       1 junction bases 0.U2330 [U]; 0.C2329+0.U2330+0.C2331+0 [CUC]
       1 junction bases 0.A2356 [A]; 0.G2355+0.A2356+0.G2357 [GAG]
       4 junction bases 0.A2367+0.A2368+0.A2369+0.A2370 [AAAA]; 0.C2366+0.A2367+0.A2368+0.A2369+0.A2370+0.G2371 [CAAAAG]
Thanks for the consideration!

[xiangjun]

Message taken -- I'll get DSSR updated by early next week.

Enjoy the weekend!

Xiang-Jun

[xiangjun]

I have revised DSSR to output loops in the format you suggested. Before I make a new beta release, please have a look of the sample run on 1s72 (attached), and report back how it goes.

Note that I have kept the original list which contains all nts in a loop. The new additions for each loop fragment are listed underneath. For the two cases you mentioned, the new output looks like below:

Code: [Select]

  55 asymmetric internal loop: 7 nts; [1x2]; linked by [#168, #169]
       0.G2855+0.A2856+0.C2857+0.G2900+0.C2901+0.A2902+0.C2903 [GACGCAC]
       1 base(s) 0.A2856 [A]; 0.G2855+0.A2856+0.C2857 [GAC]
       2 base(s) 0.C2901+0.A2902 [CA]; 0.G2900+0.C2901+0.A2902+0.C2903 [GCAC]

  20 3-way junctions: 12 nts; [1x1x4]; linked by [#138, #-43, #140]
       0.C2329+0.U2330+0.C2331+0.G2355+0.A2356+0.G2357+0.C2366+0.A2367+0.A2368+0.A2369+0.A2370+0.G2371 [CUCGAGCAAAAG]
       1 base(s) 0.U2330 [U]; 0.C2329+0.U2330+0.C2331 [CUC]
       1 base(s) 0.A2356 [A]; 0.G2355+0.A2356+0.G2357 [GAG]
       4 base(s) 0.A2367+0.A2368+0.A2369+0.A2370 [AAAA]; 0.C2366+0.A2367+0.A2368+0.A2369+0.A2370+0.G2371 [CAAAAG]
Xiang-Jun

[jyvdf3asdg2]

Looks great, makes it much easier read!

Thanks

[xiangjun]

Please download and play DSSR beta-r12-on-20130610 -- it's likely that we've moved one step forward. As always, please let me know if you have any comments or suggestions.

Note that I've slightly refined the format to make the bulges, internal loops, and helical junctions more explicit in the delineated output. Further changes are possible in future releases of DSSR, based on user feedback and other considerations.

Xiang-Jun