RNAvista Input

Help

General information
How to use RNAvista
Application example
System requirements
Citing RNAvista

General information

The RNA secondary structure containing only canonical base pairs proved insufficient for credible determination and analysis of RNA 3D structure. Thus, reliable prediction of extended secondary structure, described by both canonical and non-canonical base pairs, is of great importance in RNA structure study.

Here, we present RNAvista, a versatile web server for reliable prediction of extended RNA secondary structure. It is an implementation of our former concept (Rybarczyk et al.) based on four-stage computation: (1) optional: canonical secondary structure is predicted from a user-provided RNA sequence (if the user provides the canonical secondary structure in dot-bracket notation at the input, this step is skipped); (2) RNA 3D model is predicted from secondary structure containing canonical base pairs only; (3) the obtained 3D model is used to derive the extended secondary structure; (4) the extended secondary structure is drawn, non-canonical base pairs are classified and annotated in the graphical representation. The tool effectively incorporates specialized engine versions of RNAComposer in the second stage, and RNApdbee in the third stage of the process. The general idea of RNAvista is presented on the following diagram:

How to use RNAvista

RNAvista allows to work on a single RNA molecule at a time. Input sequence length is limited up to 500 nt residues.

Step 1: defining input data

Input data consists of: task identifier, sequence, and - optionally - canonical secondary structure topology (in dot-bracket notation) or name of an application for secondary structure prediction (selected from 6 available tools). Each information is provided in a separate line. A typical input should be formatted in the following way:

line	type	contents
1^st line	obligatory	">" followed by a unique task identifier
2^nd line	obligatory	RNA sequence in the one-letter format
3^rd line	optional	canonical secondary structure topology in dot-bracket notation / application for prediction of RNA secondary structure including canonical base pairs only

Optionally, in any place of the entry, the user can insert a comment line starting from “#”.

First line indicates where task definition begins. It must be placed before the sequence and canonical secondary structure topology specification. This line is followed by major details of an input, in which user can specify:

RNA sequence, or
RNA sequence and canonical secondary structure topology in dot-bracket, or
RNA sequence and application for canonical secondary structure prediction.

If the 3rd line of the entry remains empty, CentroidFold is applied by default to predict canonical base pairs. Other incorporated prediction algorithms can be used upon user selection in the Advanced options panel.

RNAvista uses dot-bracket notation to represent RNA secondary structure topology. In this notation:

an unpaired nucleotide is represented as a dot “.”
a base pair is represented as a pair of opening (left) and closing (right) brackets, i.e. “(” and “)”.

In our system, this notation has been extended to represent secondary structure of pseudoknots, by inserting squared “[” and “]” brackets for lower-order structures, while the curly brackets “{” and “}” and angle brackets “<” and “>” are used for higher and most complicated pseudoknots.

For user convenience, the following applications that predict canonical base pairs constituting the basic topology of RNA secondary structure have been incorporated into RNAvista pipeline:

CentroidFold (default selection),
ContextFold,
CONTRAfold,
IPknot,
RNAfold,
RNAstructure.

Step 2: setting advanced options (optional)

After clicking "Show advanced options", RNAvista displays advanced options panel that allow the user to set parameters of intermediate processing steps. Here, the user can determine:

which algorithm should be used to predict canonical secondary structure (this option is diplayed if the secondary structure has not been provided at the input; one of six algorithms can be chosen;
an application used to derive base-pairs in the predicted RNA 3D structure, which should be selected from three available programs: 3DNA/DSSR (default), RNAView, MC-Annotate; 3DNA/DSSR can be selected with additional option “Analyse helices”; RNAvista extended secondary structure contains only canonical base-pairs derived from predicted 3D model, while non-canonical ones are included in a separate list and graphical representation;
a routine to resolve and encode the extended secondary structure topology, especially when handling complex RNA with pseudoknots, should be selected among four available algorithms: Hybrid (exact + random walk) (default), Dynamic Programming.

Step 3: running computation

To start processing, the user should click Run button. This results in submitted task processing by specialized engines dedicated to RNA 3D structure modeling as well as RNA extended secondary structure annotation. At the output a results page is presented including:

extended secondary structure in selected text format
canonical and/or non-canonical base pairs in selected text format
graphical image(s) of the secondary structure
associated RNA 3D structure model visualized by JSmol
3D structures of structural elements that were assembled during 3D structure prediction
log file.

If - due to many processing requests - the submitted task cannot be executed at the time of submission, user should save the URL of current submission page and try visiting this page later to see the results. Task results will expire after 14 days since the date of submission and will be removed from the system.

Application example

Here we present how the RNAvista pipeline predicted the extended secondary structure of archaeal tyrosyl-tRNA. This example molecule is a component of an archaeal tyrosyl-tRNA synthetase complexed with tRNA(Tyr) and L-tyrosine (PDB: 1J1U).

In the first step, a user should provide an input data, here a sequence of nucleotides in FASTA format, since in this case the extended secondary structure is predicted from the sequence. The table below shows such an input data for the example molecule (archaeal tyrosyl-tRNA).

user query

>example1

CCGGCGGUAGUUCAGCCUGGUAGAACGGCGGACUGUAGAUCCGCAUGUCGCUGGUUCAAAUCCGGCCCGCCGGA

A user can optionally select the Advanced options panel to choose the other algorithms used for the extended secondary structure prediction then default.

First, the application dedicated to predict canonical secondary structure from the sequence of nucleotides can be changed (CentroidFold is set as default). In this example RNAstructure has been selected. This step is omitted if canonical secondary structure was provided together with the sequence of nucleotides as an input.

Second, a user can decide, which application will be used to extract the RNA secondary structure from the PDB-encoded atom coordinate data (default is 3DNA/DSSR). In this example, three runs with RNAvista has been conducted, each time different tool has been selected.

Next, a user can select a routine to resolve and encode the extended secondary structure topology, especially important when handling complex RNA with pseudoknots (default is Hybrid (exact + random walk)). In case of this example, the default selection has been left as is.

Finally, the processing can be started by clicking Run button.

In case of this example, the secondary model was predicted from sequence and compared to the reference structure. Detailed structural information, including non-canonical base pairs with classification, were taken from NDB (NDB: PR0092) and RNA STRAND (RNA STRAND: PDB_00474) databases. By this comparison, including canonical and non-canonical interactions, the PPV, TPR and MCC values were manually calculated (see Table below).

The following table shows the accuracy of secondary structure model predicted from tyrosyl-tRNA sequence (best values in bold). Here, including canonical and non-canonical interactions, the PPV, TPR and MCC values were manually calculated.

Variant I: Canonical and non-canonical base pairs	PPV	TPR	MCC
RNAwolf	0.71	0.44	0.56
MC-Fold-DP	0.41	0.33	0.37
MC-Fold	0.57	0.44	0.50
RNAvista (2D structure derived from 3D structure with RNAView)	0.94	0.74	0.83
RNAvista (2D structure derived from 3D structure with MC-Annotate)	0.97	0.77	0.86
RNAvista (2D structure derived from 3D structure with 3DNA/DSSR)	0.94	0.77	0.85
Variant II: Canonical base pairs only	PPV	TPR	MCC
RNAwolf	0.80	0.76	0.78
MC-Fold-DP	0.28	0.43	0.35
MC-Fold	0.56	0.71	0.63
strong	0.95	1.00	0.98
RNAvista (2D structure derived from 3D structure with MC-Annotate)	1.00	1.00	1.00
RNAvista (2D structure derived from 3D structure with 3DNA/DSSR)	1.00	1.00	1.00
Variant III: Non-canonical base pairs only, regardless of classification	PPV	TPR	MCC
RNAwolf	0.25	0.06	0.12
MC-Fold-DP	n/a	0	n/a
MC-Fold	0.67	0.11	0.27
strong	0.89	0.44	0.63
RNAvista (2D structure derived from 3D structure with MC-Annotate)	0.90	0.50	0.67
RNAvista (2D structure derived from 3D structure with 3DNA/DSSR)	0.82	0.50	0.64
Variant IV: Non-canonical base pairs only, classification dependent	PPV	TPR	MCC
RNAwolf	0.25	0.06	0.12
MC-Fold-DP	n/a	0	n/a
MC-Fold	0.33	0.06	0.14
strong	0.78	0.39	0.55
RNAvista (2D structure derived from 3D structure with MC-Annotate)	0.80	0.44	0.60
RNAvista (2D structure derived from 3D structure with 3DNA/DSSR)	0.55	0.33	0.43

System requirements

RNAvista is designed to work with most available web browsers. The latest versions of browsers are strongly recommended.

Operating system	Recommended browsers
Windows	Microsoft Internet Explorer (11.0 and later), Mozilla Firefox (50.0 and later), Google Chrome (60.0 and later)
Linux	Mozilla Firefox (50.0 and later), Google Chrome (60.0 and later)
macOS	Mozilla Firefox (50.0 and later), Google Chrome (60.0 and later)

Citing RNAvista

Any published work, which has made use of RNAvista should cite the following paper:

M. Antczak, M. Zablocki, T. Zok, A. Rybarczyk, J. Blazewicz, M. Szachniuk. RNAvista: a webserver to assess RNA secondary structures with non-canonical base pairs, Bioinformatics 35(1), 2019, 152-155 (doi: 10.1093/bioinformatics/bty609).