QC annotations
RNAcentral provides a variety of quality checks for all sequences. Many of these checks are based on Rfam. Rfam is a database of functional non-coding RNA families represented by multiple sequence alignments and consensus secondary structures. The sequence and structural information is used to build Infernal covariance models, which can be used to find new instances of RNA families and annotate genomes with non-coding RNAs.
RNAcentral annotates all sequences with Rfam models for each release. Rfam classification provides additional context to sequences with few annotations and helps identify potential problems, for example, sequences which are likely contamination.
In addition, three tools are used to detect potential protein-coding regions (see Possible ORFs).
Types of quality control
1. Incomplete sequences
When an RNAcentral sequence matches only a part of the Rfam covariance model, the sequence is flagged as incomplete.
Examples
These sequences can be browsed by searching for
qc_warning:"incomplete_sequence".
2. Potential contamination
For example, when an Eukaryotic sequence matches an Rfam family that is only found in Bacteria, this could indicate bacterial contamination or taxonomic misclassification.
Examples
Sequences of this type can be browsed with
qc_warning:"possible_contamination".
3. Missing Rfam hits
The majority of RNAcentral sequences annotated as rRNA or tRNA match the corresponding Rfam families. However, some sequences do not match the expected Rfam families which could mean that either the sequence has an incorrect RNA type or that the Rfam model needs to be updated.
Examples
Browse all sequences with missing matches by searching for
qc_warning:"missing_rfam_match".
4. Possible ORFs
Very few sequences contain open reading frames, however, it is worth noting which do as this may influence the function of the ncRNA. Three tools are used to detect them, and each has its own warning:
CPAT (v3.0.4, default options) detects possible open reading frames and is run on all human, fly, mouse, and zebrafish sequences (
qc_warning:"possible_orf").tcode (EMBOSS v6.6.0 with default options) applies the Fickett TESTCODE statistic and is run across all organisms on sequences longer than 200 nt (
qc_warning:"possible_orf_tcode").stopFree, an in-house Rust implementation of the tool described in Evaluating computational tools for protein-coding sequence detection: Are they up to the task? (v0.2.4), measures the longest stop-codon-free run in each reading frame and is run on all sequences (
qc_warning:"possible_orf_stopfree").
Examples
Quality control table
The QC status section of each sequence page has a summary table for assessing the status of a sequence at a glance. The table reports the results of coding potential, contamination, and Rfam quality-control checks, using a green tick to indicate a passing result and a red cross to indicate a failed result.
For the coding potential row, you may also see an amber status. This row can be expanded to view the results of the three coding potential tests currently used: CPAT, tcode, and stopFree. Each individual test result is displayed as either pass (green), ORF predicted (red), or not tested (grey).
When all three tests agree, the overall coding potential status is shown as either green or red. However, when the tests produce conflicting results, the overall status is shown in amber to indicate disagreement between the methods.
Example of sequence with an ORF called by all three coding potential tools: Human circRNA
Example of sequence with a disagreement between methods: Human lncRNA
Why some sequences do not match any Rfam families
There are several possible reasons:
not all RNA types are represented in Rfam
For example, piRNAs or mature miRNAs are too short to be accurately modelled in Rfam. Conversely, lncRNAs tend to be too long and poorly conserved, although Rfam includes several conserved lncRNA domains.
this family may not yet exist in Rfam
the Rfam model needs to be updated to include this sequence
If you would like to suggest a new Rfam family or report an error, please get in touch using the Feedback button found at the top right of every RNAcentral page.