# Sequence search
The RNAcentral [sequence similarity search](https://rnacentral.org/sequence-search) enables
searches against a comprehensive collection of non-coding RNA sequences from a consortium of
[RNA databases](expert-databases.md).
```{raw} html
```
(nhmmer)=
## Powered by nhmmer
The search is powered by the
[nhmmer](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3777106/) software which is more sensitive
than *blastn* but is comparable in speed. The command executed by the sequence search is as
follows:
```bash
nhmmer --qfasta --tformat fasta -o -T 0 --F3 0.02 --rna --watson --cpu 4 -Z
```
Where:
* **--F3 0.02** is only used when the sequence length is less than 50 nucleotides.
* **-Z \** is calculated according to the size of the database.
Each parameter is described in the
[HMMER User Guide](https://eddylab.org/software/hmmer/Userguide.pdf).
(query-length)=
## Query length
Queries must be between **10 and 7,000 nucleotides** long.
(exact-matches)=
## Exact sequence matches
Whenever a sequence is entered in the search input box, the query is compared with all RNAcentral
sequences and if there is an **exact match**, the links to the entries matching the query are
displayed in a green "Identical match" box. This is very quick because only identical matches are
considered. To see all similar sequences, just click Submit.
(rrna)=
## Searching for ribosomal RNAs
Over 50% of RNAcentral sequences are ribosomal RNAs (rRNAs). The abundance and high conservation of
rRNA sequences makes it difficult to perform sequence similarity searches, as such searches are
expected to match a large number of sequences and can take a long time to complete.
To get around this, the sequence similarity searches are performed against a subset of ~100,000
rRNA sequences from Ensembl, FlyBase, HGNC, MGI, PDBe, PomBase, RDP, RefSeq, RGD, SGD, TAIR, and
WormBase.
(batch-queries)=
## Batch queries
The web interface accepts up to **50 sequences** per submission, so use the
[API](sequence-search-api.md) for larger batches.
(rfam)=
## Rfam classification
In addition to nhmmer searches against RNAcentral, every query is automatically compared with the
[Rfam](https://rfam.org) library of RNA families. The searches are done using the
[Infernal](https://eddylab.org/infernal/) cmscan program coupled with a
[post-processing step](https://github.com/nawrockie/cmsearch_tblout_deoverlap) that removes hits
overlapping Rfam families from the same clan (a clan is a set of homologous families, for example
LSU_rRNA_archaea, LSU_rRNA_bacteria and LSU_rRNA_eukarya). The [Rfam website](https://rfam.org)
and the [EBI cmscan service](https://www.ebi.ac.uk/jdispatcher/rna/infernal_cmscan) report these
redundant hits.
(r2dt)=
## Secondary structure
Every search also sends your query sequence to [R2DT](r2dt.md), which selects the best-matching
template from its library of ~4,000 RNA structures, folds the query into that template's
secondary structure (2D) and draws it in the template's standard layout, so the query can be
compared with known structures of the same type. The diagram appears in the
**Secondary structure** section of the results page, and **View** opens it in the R2DT viewer,
where it can be explored and downloaded. If the query does not match any R2DT template, no
diagram is shown.
(number)=
## Number of similar sequences
Although the number of similar sequences can reach tens of thousands, for performance reasons, only
the **top 1000 results** will be shown in each search.
(sequence-search-download)=
## Download search results
The search results can be downloaded as a ZIP file containing:
- *datapackage.json*, with the search metadata;
- *sequences/similar-sequences.json* and *sequences/similar-sequences.txt*, with the
description, E-value, identity, query coverage, gaps and alignment of each match (the text
file also gives each match's RNAcentral ID);
- *sequences/similar-sequences.fasta*, with the aligned region of each matching sequence, named
as `URS…/start-end`.
(stable-links)=
## How long are the search results available for?
The results will be available at the same URL for **up to 7 days**. After this period, **all data
and metadata are deleted**.