X3DNA-DSSR: a resource for structural bioinformatics of nucleic acids
(An NIGMS National Resource supported by NIH grant R24GM153869)

Recently, PDB begins to release atomic coordinates of large (ribosomal) structures in mmCIF format. For nucleic-acid-containing structures, the largest one so far is 4v4g, the crystal structure of five 70S ribosomes from Escherichia coli in complex with protein Y. It is assembled from ten PDB entries (1voq, 1vor, 1vos, 1vou, 1vov, 1vow, 1vox, 1voy, 1voz, 1vp0), consisting of 22,345 nucleotides, and a total of 717,805 atoms.

This humongous structure poses no problems to DSSR at all, as shown below.

Command: x3dna-dssr -i=4v4g.cif -o=4v4g.out
Processing file '4v4g.cif' [4v4g]

total number of base pairs: 9277
total number of multiplets: 918
total number of helices: 1099
total number of stems: 1221
total number of isolated WC/wobble pairs: 603
total number of atom-base stacking interactions: 1736
total number of hairpin loops: 504
total number of bulges: 170
total number of internal loops: 775
total number of junctions: 214
total number of non-loop single-stranded segments: 429
total number of kissing loops: 5
total number of A-minor (type I and II) motifs: 100
total number of ribose zippers: 58 (1159)
total number of kink turns: 39

Time used: 00:00:10:45

It took less than 11 minutes to run on an iMac (and nearly 14 minutes on a Ubuntu Linux machine). Given the

With great pleasure, I read the following annoancement from Rajendra Kumar on the 3DNA Forum:

Re: do_x3dna: a tool to analyze DNA/RNA in molecular dynamics trajectories 
« Reply #1 on: Today at 10:53:31 AM »

Hello,

I have now made a new website for do_x3dna
(http://rjdkmr.github.io/do_x3dna). This website contains detailed
documentation for do_x3dna program and Python APIs.

Documentation for Python API is now available
(http://rjdkmr.github.io/do_x3dna/apidoc.html).

Few tutorials about the Python APIs are also now available
(http://rjdkmr.github.io/do_x3dna/tutorial.html).

Thanks.

With best regards,
Rajendra

Browsing through the do_x3dna website, I am impressed by the extensive documentation and tutorial. Clearly, do_x3dna has pushed the boundaries (in applicability and documentation) of the x3dna_ensemble Ruby script distributed with 3DNA v2.1.

As noted in GitHub page, do_x3dna has been developed to analyze fluctuations in DNA or RNA structures in molecular dynamics (MD) trajectories. It can be used for GROMACS MD trajectories, as well as those from NAMD and AMBER. It leaves no doubt that do_x3dna will boost 3DNA’s applications in the increasingly active field of DNA/RNA MD simulations.

From early on, 3DNA and DSSR have native support of modified nucleotides. The currently distributed baselist.dat file with 3DNA contains over 700 entries. As of v1.1.4-2014aug09, a new section has been added to DSSR to list explicitly the modified nucleotides in an analyzed structure.

Using the 76-nucleotide long yeast phenylalanine tRNA (1ehz) as an example, the pertinent section in DSSR output is as below.

List of 11 types of 14 modified nucleotides
      nt    count  list
   1 1MA-a    1    A.1MA58
   2 2MG-g    1    A.2MG10
   3 5MC-c    2    A.5MC40,A.5MC49
   4 5MU-t    1    A.5MU54
   5 7MG-g    1    A.7MG46
   6 H2U-u    2    A.H2U16,A.H2U17
   7 M2G-g    1    A.M2G26
   8 OMC-c    1    A.OMC32
   9 OMG-g    1    A.OMG34
  10 PSU-P    2    A.PSU39,A.PSU55
  11 YYG-g    1    A.YYG37

So 1ehz has 14 modified nucleotides of 11 different type, as listed in the following rows after the header line. The meaning of each column should be obvious. For example, the third row means that 5MC (5-methylcytidine, abbreviated as 'c' in 1-letter code) occurs twice, identified as A.5MC40 and A.5MC49, respectively.

With the 3-letter id, one can search the RCSB ligand database for more information about a specified modified nucleotide. The URL would be like this, using pseudouridine (PSU) as an example, http://www.rcsb.org/pdb/ligand/ligandsummary.do?hetId=PSU.

It is hoped that the newly added section, put at the very top of DSSR output, will draw more attention to modified nucleotides.

From v1.1.3-2014jun18, DSSR has an additional output of RNA secondary structures in BPSEQ format. A sample file for PDB entry 1msy is shown below.

Filename: dssr-2ndstrs.bpseq
Organism: DSSR-derived secondary structure [1msy]
Accession Number: DSSR v1.1.4-2014aug09 (xiangjun@x3dna.org)
Citation: Please cite 3DNA/DSSR (see http://home.x3dna.org)
    1 U     0 # name=A.U2647
    2 G    26 # name=A.G2648, pairedNt=A.U2672
    3 C    25 # name=A.C2649, pairedNt=A.G2671
    4 U    24 # name=A.U2650, pairedNt=A.A2670
    5 C    23 # name=A.C2651, pairedNt=A.G2669
    6 C    22 # name=A.C2652, pairedNt=A.G2668
    7 U     0 # name=A.U2653
    8 A     0 # name=A.A2654
    9 G     0 # name=A.G2655
   10 U     0 # name=A.U2656
   11 A     0 # name=A.A2657
   12 C    17 # name=A.C2658, pairedNt=A.G2663
   13 G     0 # name=A.G2659
   14 U     0 # name=A.U2660
   15 A     0 # name=A.A2661
   16 A     0 # name=A.A2662
   17 G    12 # name=A.G2663, pairedNt=A.C2658
   18 G     0 # name=A.G2664
   19 A     0 # name=A.A2665
   20 C     0 # name=A.C2666
   21 C     0 # name=A.C2667
   22 G     6 # name=A.G2668, pairedNt=A.C2652
   23 G     5 # name=A.G2669, pairedNt=A.C2651
   24 A     4 # name=A.A2670, pairedNt=A.U2650
   25 G     3 # name=A.G2671, pairedNt=A.C2649
   26 U     2 # name=A.U2672, pairedNt=A.G2648
   27 G     0 # name=A.G2673

Based on online sources, BPSEQ has originated from the Comparative RNA Web site developed by the Gutell lab. CRW files contain four header lines, describing the file name, organism, accession number, and a general remark. Thereafter, there is one line per base in the molecule, listing the position of the base (starting from 1), the one-letter base name (A,C,G,U etc), and the position number of the base to which it is paired. If the base is unpaired, zero (0) is put in the third column. In the above sample BPSEQ file derived from DSSR, detailed information about the base and its paired base (if any) comes after the # symbol.

Compared to dot-bracket notation (dbn) and connect-table (.ct) format, BPSEQ is simpler but less expressive. Nevertheless, the format is well-supported in bioinformatic tools on RNA secondary structures. It only seems fitting that DSSR now produces secondary structures in .bpseq (with default file name dssr-2ndstrs.bpseq), in addition to .dbn and .ct. Technically, adding the BPSEQ output to DSSR is trivial given the infrastructure already in place.

From early on, DSSR-derived RNA secondary structures in dot-bracket notation (dbn) have taken pseudoknots into consideration. Nevertheless, in DSSR releases prior to v1.1.3-2014jun18, the dbn output had been simplified to the first level only, with matched []s, even for RNA structures with high-order pseudoknots. RNA pseudoknot is a (relatively) complicated issue, and I’d planned to put off the topic until DSSR is well-established.

In early May, I noticed the Antczak et al. article RNApdbee—a webserver to derive secondary structures from pdb files of knotted and unknotted RNAs. I was delighted to read the following citation:

In order to facilitate a more comprehensive study, the webserver integrates the functionality of RNAView, MC-Annotate and 3DNA/DSSR, being the most common tools used for automated identification and classification of RNA base pairs.

Even before any paper on DSSR has been published, the software has already be ranked in the top three for the identification and classification of RNA base pairs! Well familiar with RNAView and MC-Annotate, I am glad to see DSSR is now listed on a par with them. Note that DSSR has far more functionality than just identifying and classifying RNA base pairs.

Further down the RNApdbee paper, especially in Figure 2, I found the following remarks regarding DSSR’s capability on RNA structures with high-order pseudoknot.

An arc diagram to represent the secondary structure of 1DDY (chain A) generated by R-CHIE upon the dot-bracket notation. Arcs of the same colour define a paired region. Crossing arcs reflect a conflict observed between the corresponding regions. (a) RNApdbee recognizes pseudoknots of the first (dark green) and second (navy blue) order. (b) 3DNA/DSSR improperly classifies base pairs (within residues in red) and the structure is recognized as the first-order pseudoknot.

The above citation and the question Higher-order pseudoknots in DP output (from Jan Hajic, Charles University in Prague) on the 3DNA Forum prompted me to further refine DSSR’s algorithm for deriving secondary structures of RNA with high-order pseudoknots. The DSSR v1.1.3-2014jun18 release made this revised functionality explicit. For the above cited PDB entry 1ddy, the relevant output of running DSSR on it would be:

Running command: "x3dna-dssr -i=1ddy.pdb"

****************************************************************************
This structure contains 2-order pseudoknot(s)

****************************************************************************
Secondary structures in dot-bracket notation (dbn) as a whole and per chain
>1ddy nts=140 [whole]
GGAACCGGUGCGCAUAACCACCUCAGUGCGAGCAA&GGAACCGGUGCGCAUAACCACCUCAGUGCGAGCAA&GGAACCGGUGCGCAUAACCACCUCAGUGCGAGCAA&GGAACCGGUGCGCAUAACCACCUCAGUGCGAGCAA
......(((.{[[....[[)))...].].}.]]..&......(((.{[[....[[)))...].].}.]]..&......(((.{[[....[[)))...].].}.]]..&......(((.{[[....[[)))...].].}.]]..
>1ddy-A #1 nts=35 [chain] RNA
GGAACCGGUGCGCAUAACCACCUCAGUGCGAGCAA
......(((.{[[....[[)))...].].}.]]..
>1ddy-C #2 nts=35 [chain] RNA
GGAACCGGUGCGCAUAACCACCUCAGUGCGAGCAA
......(((.{[[....[[)))...].].}.]]..
>1ddy-E #3 nts=35 [chain] RNA
GGAACCGGUGCGCAUAACCACCUCAGUGCGAGCAA
......(((.{[[....[[)))...].].}.]]..
>1ddy-G #4 nts=35 [chain] RNA
GGAACCGGUGCGCAUAACCACCUCAGUGCGAGCAA
......(((.{[[....[[)))...].].}.]]..

Note that the whole 1ddy entry contains four RNA chains (A, C, E, and G), and DSSR can handle each properly. So at least from DSSR v1.1.3-2014jun18, the following statement is no longer valid:

3DNA/DSSR improperly classifies base pairs (within residues in red) and the structure is recognized as the first-order pseudoknot.

A closely related issue is knot removal, a topic nicely summarized by Smit et al. in their publication From knotted to nested RNA structures: A variety of computational methods for pseudoknot removal. While not explicitly documented, the --nested (abbreviated to --nest) option has been available since DSSR v1.1.3-2014jun18. This option was first mentioned in the release note of DSSR v1.1.4-2014aug09. Again, using PDB entry 1ddy as an example, the relevant output of running DSSR with option --nested is as follows:

Running command: "x3dna-dssr -i=1ddy.pdb --nested"

****************************************************************************
This structure contains 2-order pseudoknot(s)
   o You've chosen to remove pseudo-knots, leaving only nested pairs

****************************************************************************
Secondary structures in dot-bracket notation (dbn) as a whole and per chain
>1ddy nts=140 [whole]
GGAACCGGUGCGCAUAACCACCUCAGUGCGAGCAA&GGAACCGGUGCGCAUAACCACCUCAGUGCGAGCAA&GGAACCGGUGCGCAUAACCACCUCAGUGCGAGCAA&GGAACCGGUGCGCAUAACCACCUCAGUGCGAGCAA
......(((..........))).............&......(((..........))).............&......(((..........))).............&......(((..........))).............
>1ddy-A #1 nts=35 [chain] RNA
GGAACCGGUGCGCAUAACCACCUCAGUGCGAGCAA
......(((..........))).............
>1ddy-C #2 nts=35 [chain] RNA
GGAACCGGUGCGCAUAACCACCUCAGUGCGAGCAA
......(((..........))).............
>1ddy-E #3 nts=35 [chain] RNA
GGAACCGGUGCGCAUAACCACCUCAGUGCGAGCAA
......(((..........))).............
>1ddy-G #4 nts=35 [chain] RNA
GGAACCGGUGCGCAUAACCACCUCAGUGCGAGCAA
......(((..........))).............

H-bonding interactions are crucial for defining RNA secondary and tertiary structures. DSSR/3DNA contains a geometrically based algorithm for identifying H-bonds in nucleic-acid or protein structures given in .pdb or .cif format. Over the years, the method has been continuously refined, and it has served its purpose quite well. As of v1.1.1-2014apr11, this functionality is directly available from DSSR thorough the --get-hbonds option.

The output for 1msy, which contains a GUAA tetraloop mutant of Sarcin/Ricin domain from E. Coli 23 S rRNA, is listed below. The first line gives the header (# H-bonds in '1msy.pdb' identified by DSSR ...). The second line provides the total number of H-bonds (40) identified in the structure. Afterwards, each line consists of 8 space-delimited columns used to characterize a specific H-bond. Using the first one (#1) as an example, the meaning of each of the 8 columns is:

1. The serial number (15), as denoted in the .pdb or .cif file, of the first atom of the H-bond.
2. The serial number (578) of the second H-bond atom.
3. The H-bond index (#1), from 1 to the total number of H-bonds.
4. A one-letter symbol showing the atom-pair type (p) of the H-bond. It is ‘p’ for a donor-acceptor atom pair; ‘o’ for a donor/acceptor (such as the 2′-hydorxyl oxygen) with any other atom; ‘x’ for a donor-donor or acceptor-acceptor pair (as in #17); ‘?’ if the donor/acceptor status is unknown for any H-bond atom.
5. Distance in Å between donor/acceptor atoms (2.768).
6. Elemental symbols of the two atoms involved in the H-bond (O/N).
7. Identifier of the first H-bond atom (O4@A.U2647).
8. Identifier of the second H-bond atom (N1@A.G2673).

Command: x3dna-dssr -i=1msy.pdb --get-hbonds –o=1msy-hbonds.txt

# H-bonds in '1msy.pdb' identified by 3DNA version 3 (xiangjun@x3dna.org)
40
   15   578  #1     p    2.768 O:N O4@A.U2647 N1@A.G2673
   35   555  #2     p    2.776 O:N O6@A.G2648 N3@A.U2672
   36   554  #3     p    2.826 N:O N1@A.G2648 O2@A.U2672
   55   537  #4     p    2.965 O:N O2@A.C2649 N2@A.G2671
   56   535  #5     p    2.836 N:N N3@A.C2649 N1@A.G2671
   58   534  #6     p    2.769 N:O N4@A.C2649 O6@A.G2671
   76   513  #7     p    2.806 N:N N3@A.U2650 N1@A.A2670
   78   512  #8     p    3.129 O:N O4@A.U2650 N6@A.A2670
   95   492  #9     p    2.703 O:N O2@A.C2651 N2@A.G2669
   96   490  #10    p    2.853 N:N N3@A.C2651 N1@A.G2669
   98   489  #11    p    2.987 N:O N4@A.C2651 O6@A.G2669
  115   466  #12    p    2.817 O:N O2@A.C2652 N2@A.G2668
  116   464  #13    p    2.907 N:N N3@A.C2652 N1@A.G2668
  118   463  #14    p    2.897 N:O N4@A.C2652 O6@A.G2668
  123   151  #15    o    2.622 O:O OP2@A.U2653 O2'@A.A2654
  135   443  #16    p    2.898 O:N O2@A.U2653 N4@A.C2667
  147   192  #17    x    3.054 O:O O4'@A.A2654 O4'@A.U2656
  158   408  #18    p    2.960 N:O N6@A.A2654 OP2@A.C2666
  173   188  #19    o    2.923 O:O O2'@A.G2655 OP2@A.U2656
  173   378  #20    o    3.093 O:O O2'@A.G2655 O6@A.G2664
  173   379  #21    o    3.343 O:N O2'@A.G2655 N1@A.G2664
  181   386  #22    p    2.768 N:O N1@A.G2655 OP2@A.A2665
  183   203  #23    p    2.754 N:O N2@A.G2655 O4@A.U2656
  183   387  #24    p    2.887 N:O N2@A.G2655 O5'@A.A2665
  188   379  #25    p    3.044 O:N OP2@A.U2656 N1@A.G2664
  188   381  #26    p    2.944 O:N OP2@A.U2656 N2@A.G2664
  200   401  #27    p    3.122 O:N O2@A.U2656 N6@A.A2665
  201   398  #28    p    2.759 N:N N3@A.U2656 N7@A.A2665
  220   381  #29    p    3.035 N:N N7@A.A2657 N2@A.G2664
  223   371  #30    o    2.963 N:O N6@A.A2657 O2'@A.G2664
  223   382  #31    p    3.039 N:N N6@A.A2657 N3@A.G2664
  242   358  #32    p    2.821 O:N O2@A.C2658 N2@A.G2663
  243   356  #33    p    2.890 N:N N3@A.C2658 N1@A.G2663
  245   355  #34    p    2.887 N:O N4@A.C2658 O6@A.G2663
  258   305  #35    o    2.604 O:N O2'@A.G2659 N7@A.A2661
  258   308  #36    o    3.264 O:N O2'@A.G2659 N6@A.A2661
  268   315  #37    p    2.973 N:O N2@A.G2659 OP2@A.A2662
  268   327  #38    p    2.864 N:N N2@A.G2659 N7@A.A2662
  371   390  #39    o    2.751 O:O O2'@A.G2664 O4'@A.A2665
  550   566  #40    o    3.372 O:O O2'@A.U2672 O4'@A.G2673

In its default settings, DSSR detects 117 H-bonds for 1ehz (yeast phenylalanine tRNA), and 5,809 for 1jj2 (the H. marismortui large ribosomal subunit). Note that the program can identify H-bonds not only in RNA and DNA, but also in proteins, or their complexes. By default, however, DSSR only reports H-bonds within nucleic acids. As shown above, it is trivial to run DSSR with the --get-hbonds option to get all H-bonds in a given structure, and the plain text output is straightforward to work on.

While there exist dedicated tools for finding H-bonds, such as HBPLUS or HBexplore, DSSR may well be sufficient to fulfill most practical needs. If you notice any weird behaviors with this H-bond finding functionality, please let me know. I strive to address reported issues promptly, to the extent practical. At the very least, I should be able to explain why the program is working the way it does.

From the very first release up until recently, the DSSR distribution had included two executables for Windows: one version was compiled on MinGW/MSYS, and the other on Cygwin. The executables are supposed to be run under the corresponding shells of the two environments respectively.

Since DSSR is a simple self-contained command-line tool, the MinGW/MSYS version also works directly under the Command Prompt of native Windows. So Windows users had the following three options to use DSSR:

• Download the MinGW/MSYS version to run it under the Command Prompt of native Windows. No need to install MinGW/MSYS.
• Download the MinGW/MSYS version to run it under the MinGW/MSYS environment, which must be installed separately.
• Download the Cygwin version to run it under the Cygwin environment, which must be installed separately.

Over times, I have observed some confusions among DSSR users as to which of the two executables to use on Windows. Luckily, I noticed by chance recently that the DSSR executable compiled under MinGW/MSYS runs just fine in the Cygwin shell. So as of v1.1.0-2014apr09, the DSSR distribution contains only one executable for Windows: compiled under MinGW/MSYS on 32-bit Windows XP, the same DSSR executable runs under the Command Prompt of native Windows, MinGW/MSYS, and Cygwin, either on a 32-bit or 64-bit Windows (XP, Vista, 7 or 8) machine.

A size fits all: I no longer need to provide two compiled versions of DSSR for Windows, and users have just one executable to download (no more space for confusions).

Note added on 2024-11-25: DSSR is distributed by the CTV (Columbia Technology Ventures). See https://home.x3dna.org

In addition to VARNA, the draw program in the RNAstructure package from the Mathews Laboratory can also be used to depict DSSR-derived RNA secondary structures in connect table (.ct) format. The draw program produces images in PostScript (or svg) format, in different styles from those generated by VARNA. Given below are a couple of examples on how to connect DSSR with draw.

The secondary structure of the PDB entry 1msy in DSSR-derived .ct file is as below:

   27 DSSR-derived secondary structure in '1msy'
    1 U     0     2     0  2647
    2 G     1     3    26  2648
    3 C     2     4    25  2649
    4 U     3     5    24  2650
    5 C     4     6    23  2651
    6 C     5     7    22  2652
    7 U     6     8     0  2653
    8 A     7     9     0  2654
    9 G     8    10     0  2655
   10 U     9    11     0  2656
   11 A    10    12     0  2657
   12 C    11    13    17  2658
   13 G    12    14     0  2659
   14 U    13    15     0  2660
   15 A    14    16     0  2661
   16 A    15    17     0  2662
   17 G    16    18    12  2663
   18 G    17    19     0  2664
   19 A    18    20     0  2665
   20 C    19    21     0  2666
   21 C    20    22     0  2667
   22 G    21    23     6  2668
   23 G    22    24     5  2669
   24 A    23    25     4  2670
   25 G    24    26     3  2671
   26 U    25    27     2  2672
   27 G    26     0     0  2673

Let the DSSR-derived .ct file for 1msy be named 1msy.ct, the following two draw-command runs will produce the secondary structure in PostScript (1msy.eps) and svg (1msy.svg) respectively.

draw 1msy.ct 1msy.eps
draw 1msy.ct 1msy.svg --svg -n 1

The PDB entry 1ehz (yeast phenylalanine tRNA) has a pseudo knot, so the draw program will create a ‘circularized’ structure as shown below:

Note the following two caveats:

• The draw program does not like the extra information starting with # after the 6 data columns of a .ct file produced by DSSR (by default). The program hangs at “Determining structure information…”.
• For the DSSR-derived .ct file for the Dickerson DNA dodecamer, draw aborts with “Drawing structure 1…Segmentation fault: 11”.

As of v1.0.3-2014mar09, DSSR has a decent user manual in PDF! Currently of 45 pages long, the DSSR manual contains everything a typical user needs to know to get started using the program effectively. The contents the manual are listed below.

Table of Contents

List of Figures

Introduction

Download and installation

Usages
  Command-line help
  Default run on PDB entry 1msy – detailed explanations
    Summary section
    List of base pairs
    List of multiplets
    List of helices
    List of stems
    List of lone canonical pairs
    List of various loops
    List of single-stranded fragments
    Secondary structure in dot-bracket notation
    List of backbone torsion angles and suite names
  Default run on PDB entry 1ehz (tRNAPhe) – summary notes
    Brief summary
    Specific features
  Default run on PDB entry 1jj2 – four auto-checked motifs
    Kissing loops
    A-minor (types I and II) motifs
    Ribose zippers
    Kink turns
  The --more option
    Extra parameters for base pairs
    Extra parameters for helices/stems
  The –-non-pair option
  The –-u-turn option
  The --po4 option
  The –-long-idstr option

Frequently asked questions
  How to cite DSSR?
  Does DSSR work for DNA?
  Does DSSR detect RNA tertiary interactions?

Revision history

Acknowledgements

References

With the User Manual available, I feel confident to claim that DSSR is now mature, stable, ready for real world applications. While only time would tell, I have no doubt that DSSR will become an essential tool in RNA structural bioinformatics.