SeqFetcher

Use SeqFetcher (/ncbi-fetcher) to download nucleotide or protein records from NCBI by sequence accession, assembly accession or sequence UID. FASTA and metadata let you verify what was retrieved before continuing.

Prepare identifiers

Paste IDs or upload UTF-8 .txt, .csv or .tsv containing IDs only, without a header. Separate IDs with newlines, commas, spaces or semicolons. Duplicate input IDs are removed.

NC_012920.1
NM_000546.6
NM_001126112.3

These are nucleotide retrieval examples, not a bacterial CRISPR dataset. Use sequence records corresponding to your study samples. A version suffix such as .1 identifies the requested record version.

Select Nucleotide for DNA/RNA or Protein for amino-acid records such as NP_000537.3. Numeric UIDs must belong to that database. For whole-genome assemblies, select Nucleotide and supply a GCA_ or GCF_ accession, for example GCF_000339015.1. Assembly IDs are not accepted in Protein mode. Gene, Taxonomy, BioProject and SRA IDs are not supported inputs.

Retrieve sequences

1. Select the paste or upload input tab and enter IDs.
2. Check the unique-ID count and any validation message.
3. Select the database; keep nucleotide and protein requests in separate runs.
4. Choose 5 · standard under IDs per request, or 1 · large sequences for large genomes. If the list contains an assembly accession, requests automatically use batch size 1.
5. Select Commit and follow result statuses.
6. Check organism, description and length against your sample list.

Stop after this batch requests cancellation; the active batch may finish and later IDs remain pending. Resume pending submits pending IDs as a new job. Retry failed submits all failed IDs as a new job, independently of row selection. The displayed result is replaced by the new job result; earlier successes remain accessible through the previous private link. Save that link before retrying or starting another run.

Read results

Use the row statuses to follow retrieval. The downloaded run report records requested, succeeded, failed and pending counts; selected records can resolve to fewer distinct accessions.

Item	Meaning
Requested	Unique input IDs in this run.
Retrieved	Input IDs resolved successfully.
Failed	IDs that could not be retrieved; inspect the error.
Pending	IDs not completed, for example after stopping or a service interruption.
Unique sequences	Distinct retrieved accessions. Different IDs can resolve to one accession, so this can be lower than the retrieved-record count. This is not similarity clustering.
Length	Nucleotide or amino-acid residues, depending on the database.

Search by ID, accession, organism or description, and use the status filter to locate unresolved records. A preview shows the FASTA header and first 2,000 characters; downloads contain the full sequence. For sequence-accession rows, View on NCBI opens the source record. For an assembly, use its GCA_/GCF_ accession to locate the corresponding NCBI assembly record.

Successful retrieval does not establish that the record is the intended strain, a complete genome or suitable CRISPR input. Verify metadata before analysis.

Read an assembly result

An assembly appears as one accession row, but its FASTA can contain many chromosome/contig records. The reported length is the sum of their lengths. The downloader preserves separate FASTA headers and does not join contigs into one artificial sequence. Thus a count of one exported assembly is not necessarily one FASTA record.

The preview still shows only the first 2,000 characters; use the full FASTA download to inspect every contig. Check the assembly accession/version and metadata. An assembly package must match the requested version and its reported total length; incomplete or inconsistent downloads are reported as failures.

Select and download

Retrieved rows are selected automatically. Row checkboxes change the export selection. The header checkbox affects visible retrieved rows only; previously selected hidden rows can remain selected. Check the selected-record count before export. Searching alone does not limit downloads to visible rows.

Download	Contents / use
FASTA	Full sequences for selected records, deduplicated by accession, in one file.
Metadata TSV	Selected input IDs, accessions, organism, description, length and source metadata. Reconcile it with your sample list.
Download ZIP	Combined and individual FASTA, metadata TSV/JSON, run report, and failed/pending ID lists.
Download failed IDs	Unresolved IDs to check or retry later.
Preview Download	Complete FASTA for that accession, including all component contigs for an assembly.

Copy the private results link after submission. Retrieval continues on the server, and results can be restored through My Jobs for seven days after the task ends. Download FASTA/ZIP for longer storage. For several bacterial strains, preserve separate sample identities when continuing to CRISPRPrep; CRISPRPrep treats each uploaded FASTA file as one sample. Use individual FASTA files from the ZIP for distinct strains, or use its NCBI accession route, which groups retrieved contigs by requested ID. Protein FASTA cannot be used for genome CRISPR detection.

Limits and recovery

A job accepts up to 500 unique IDs and 16,000 characters of submitted ID text; ID files are limited to 1 MB. Each request contains at most 5 IDs, or 1 when assemblies are included, and returns at most 8 million residues. The saved FASTA-text budget is 25 million characters per job, including headers and line breaks. Batch size 1 does not remove the per-response limit. Assembly packages also have a 64 MiB download/unpacked-size limit.

Symptom	Action
Record not found	Check accession spelling/version and database.
WGS master has no sequence	Retrieve component records or assembly FASTA.
Batch exceeds residue limit	Set batch size to 1 and retry; a single record over the limit needs another retrieval workflow.
Temporary NCBI failure	Save the previous result link and successful outputs; retry failed/pending IDs as a new job after recovery.
Download disabled	Wait for retrieval to stop and select a retrieved record.