DNA sequencers

OpenReadout reads what DNA sequencers write, without the vendor’s software:

what	format id	page
Illumina run folders: RunInfo.xml, RunParameters.xml, InterOp metrics, BCL and CBCL base calls	illumina-run	Illumina run folders
Oxford Nanopore raw signal	pod5, fast5	POD5 and FAST5
Reads as sequencers and basecallers hand them on (Illumina, PacBio HiFi, Oxford Nanopore)	fastq, bam	FASTQ and BAM
Reference-compressed alignments (CRAM 2.0 to 3.1); bases need the reference, embedded or as a FASTA beside the file	cram	CRAM

A run folder

info reads the run’s layout and the InterOp metrics, never the base calls, so it answers in well under a second on a run of terabytes:

$ openreadout info 200624_A00834_0183_BHMTFYDRXX --json | jq '.data.tables[0].extra.totals'
{
  "yield_gbp": 56.3668,
  "percent_q30": 94.281,
  "reads_m": 1276.67,
  "reads_pf_m": 1063.38,
  "percent_pf": 83.293,
  "lanes": 2,
  "cycles": 56,
  "read_structure": "36T10B10B"
}

Table 0 is the run summary per read and lane (density, % passing filter, reads, % ≥ Q30, yield, error rate, phasing, % occupied), table 1 the metrics per lane and cycle, table 2 per tile, table 3 the Q-score histogram and table 4 the index counts. Trace 0 holds the quality by cycle with one sweep per lane, so openreadout trace RUN --sweep 1 is lane 2 and openreadout plot draws it:

$ openreadout plot '{"data": {"file": "RUN", "y": "percent_q30", "sweep": [0, 1]}}' -o q30.png

openreadout preview RUN draws the picture above. openreadout export RUN --format fastq writes gzip FASTQ per lane and read from the base calls (--rows 0-9999 for the first 10,000 clusters of each lane), and openreadout check RUN decodes every base call.

Nanopore reads

Every read is one sweep of trace 0, in picoamperes:

$ openreadout trace HG002.pod5 --sweep 3 --max-samples 5
$ openreadout preview HG002.pod5 --sweep 3

Table 0 lists the reads (channel, well, start, duration, end reason, open-pore level), and openreadout export HG002.pod5 --format parquet --trace 0 writes every read’s samples to Parquet (a sweep column numbers the reads; without --trace it writes the reads table).

Read files

info streams the first 32 MB of a FASTQ or BAM file and reports read counts, the N50, % bases ≥ Q30 and GC; it says when counts are estimated, and check reads every record.

Across runs

batch compares runs: one row of instrument, run and sample facts per run folder, or the columns of the run summary per run:

$ openreadout batch info 'runs/*/' -o runs.csv
$ openreadout batch table 'runs/*/' --set table=0 --set 'parameters=["percent_q30", "yield"]' -o q30.csv

Not read

Ion Torrent (.wells, .dat), MGI and Element Biosciences run folders, PacBio raw movie and .baz files, Sanger and fragment-analysis traces are not read yet. CRAM is not read; samtools reads it.