Downloading sequence data¶
By default adaptiseq -i <accession> resolves the accession, downloads every Run
it contains through the segmented engine, and md5-verifies each file.
adaptiseq -i PRJNA211801 # all Runs in a project
adaptiseq -i SRR7706354 # a single Run
adaptiseq -i accessions.txt # a file of accessions (batch; see batch.md)
For each Run, adaptiSeq downloads, checks the md5 against the public database,
and on mismatch retries up to three rounds before recording the result in
success.log / fail.log.
-g, --gzip — direct *.fastq.gz¶
Download FASTQ directly in gzip format.
For GSA accessions most data is already gzip, so -g is effectively always
on. For SRA/ENA/DDBJ/GEO, adaptiSeq first tries ENA's *.fastq.gz mirror; if
that is unavailable it falls back to downloading the .sra file and converting it
with fasterq-dump + pigz.
Note A pure ENA
*.fastq.gzdownload needs onlywget(pluspigzfor any conversion).sra-toolsis required only if the SRA fallback actually triggers — adaptiSeq then prints a clear "install sra-tools" message rather than failing obscurely.
-q, --fastq — convert SRA to FASTQ¶
After downloading the .sra file, decompose it into uncompressed FASTQ with
fasterq-dump.
Note
-qis particularly useful for single-cell data: it can decompose a run intoI1/R1/R2/R3, whereas a direct-gdownload may yield onlyR1/R3. Used together,-q -gdownloads the.sra, converts it, then gzips the result.
-t, --threads¶
Threads for SRA→FASTQ conversion and gzip compression (default 8).
More is not always better — fasterq-dump is IO-heavy, and too many threads
raise CPU/IO load. A ceiling around 15 is a sensible default.
-d, --database — force ENA or SRA¶
By default adaptiSeq auto-detects the best database, so -d is usually
unnecessary. Force -d sra when ENA is slow for a given file.
Note If a file is not in ENA, adaptiSeq switches to SRA automatically even with
-d ena.
-r, --protocol — ENA transport¶
For the segmented engine, ENA transport defaults to auto (HTTPS-first): it
prefers the HTTPS mirror (confirmed by a cheap per-host range probe), then native
segmented FTP, then a single stream inside the segmented engine. Pass
-r https or -r ftp to force one.
-p, --parallel — classic-engine parallelism¶
-p sets the axel connection count and applies only to the opt-in classic
engine (--engine classic). The default segmented engine already fetches each
file in multiple byte-range connections (--max-segments) and runs a batch
worker pool (-j), so -p is not needed there. See Method details.
-s, --speed — speed cap¶
Token-bucket rate limit in MB/s (default 1000), applied by the engine.
-k, --skip-md5¶
Skip md5/vdb-validate integrity checks. Remove -k and re-run to verify later.
-Q, --quiet¶
Suppress the live progress bar and non-essential messages — cleaner logs for batch/CI use.