Skip to content

MinKNOW configuration

How to configure MinKNOW for real-time BAM output compatible with ROBIN.

Overview

ROBIN expects:

  • Real-time basecalling (with methylation output as required by your ROBIN build)
  • Alignment to the same reference you pass to ROBIN
  • BAM output with sensible rollover (read-count–based, not hourly-only)
  • Clear sample (and optionally experiment) IDs for output organisation

Basecalling

Use high-accuracy basecalling with 5hmC / 5mc methylation calling where ROBIN requires it. Do not disable methylation for off-target reads if your pipeline uses those signals.

Example basecalling config (verify against your MinKNOW / kit release):

  • dna_r10.4.1_e8.2_400bps_5khz_modbases_5hmc_5mc_cg_hac_prom.cfg

Adaptive sampling

If you use adaptive sampling:

  • Use the BED file specified for your ROBIN / panel build (e.g. panel_11092024_5kb_pad.bed for classic CNS panels — confirm path in your installed resources).
  • Use the same reference as ROBIN and alignment.
  • Mode typically enrich (per your assay design).

Alignment

Produce aligned BAMs against the same reference genome files used in ROBIN (--reference).

BAM output

  • Prefer read-count rollover, not time-only hourly chunks (unsupported for real-time ROBIN processing).
  • Keep each BAM under the supported read count (see README); a common target is ~20,000–50,000 reads per file depending on workflow.
  • You may disable POD5 and FASTQ if you only need BAM.

Sample and experiment IDs

  • Set a unique sample ID per library — ROBIN uses this for output folders and to join analysis to the correct run.
  • If you use ROBIN’s Sample ID generator (MD5 from test ID and optional patient fields), use that generated sample ID as the MinKNOW sample ID when you start the run. ROBIN will then link MinKNOW outputs, the on-disk manifest, and the web app without manual renaming.
  • Experiment ID can group runs; you can point ROBIN at a subset of output directories if helpful.
  • Set run duration appropriate to your clinical or research protocol (e.g. 24 h for many CNS assays).

Run workflow

  1. Apply the settings above in MinKNOW.
  2. Start the run.
  3. Start ROBIN with the directory MinKNOW writes BAMs into (robin workflow … for Little John, or legacy robin … for Original ROBIN).
  4. Monitor progress in the CLI / NiceGUI (Little John) or legacy GUI (Original ROBIN).

Troubleshooting

  • API errors (minknow_api): install the Python minknow-api version that matches your MinKNOW install — see Installation → MinKNOW API.
  • Huge BAMs / missed processing: reduce reads per file and avoid time-only rollover.
  • Reference mismatch: alignment reference must match ROBIN’s --reference.

Next steps