escpod sidecar¶

Write the .p5s sidecar: the companion file that holds a POD5's read index, its signal batch geometry and any per-read annotations. The POD5 itself is never written — raw sequencer output stays byte-identical and checksummable, and deleting an annotation is editing or deleting the sidecar.

escpod sidecar <COMMAND> [OPTIONS]

Commands¶

One operation, two entry points¶

Both commands write the same file through the same writers, with the same binding to the POD5 (by file identifier and size, so a stale or misplaced sidecar fails loudly) and the same handling of a directory's collection sidecar. sidecar annotate builds the read index when the sidecar does not have one yet, so sidecar index is the labels-free form of the same operation: reach for it to pay the cost of the index and the geometry once, on a networked node, before submitting jobs against a large file.

Each command touches only its own columns. sidecar index keeps annotations and scores, and sidecar annotate keeps the index and other annotations.

escpod sidecar index run1/pod5/                        # caches only
escpod sidecar annotate -a demux.csv run1/pod5/        # labels (and the index, if missing)
escpod sidecar annotate run1/pod5/ \
    --from-table calls.csv.gz --column barcode=predicted_barcode

Deprecated spellings¶

escpod index and escpod annotate were the top-level spellings before the group existed. They still work, with the same arguments and behaviour: each is hidden from --help, logs one warning on stderr naming the new spelling, and forwards to the command above. They will be removed in a later release, so switch scripts now.

See also¶

• The .p5s Sidecar — the on-disk specification.
• demux — demux --annotate writes the sidecar directly from the fused pipeline, and demux split --sidecar reads it back.