Branches

The --branch option (-b) lets you run a step under a different name so its outputs go into a separate directory alongside the default outputs. This is useful when you want to try different settings on the same data without overwriting existing results.

How branches work

When you set --branch NAME, the step writes its outputs to {out}/{sample}/{step}_{NAME}/ instead of {out}/{sample}/{step}/. For example:

seismic filter --branch strict --min-finfo-read 0.99 out/sample-1/idmut/ref-1

writes to out/sample-1/filter_strict/ref-1/, leaving the default out/sample-1/filter/ref-1/ untouched.

Downstream steps can then run on either branch independently:

seismic cluster out/sample-1/filter_strict/ref-1
seismic cluster out/sample-1/filter/ref-1

Which steps support --branch

• seismic demult

• seismic align

• seismic idmut

• seismic filter

• seismic cluster

• seismic fold

• seismic filterscan

• seismic clusterscan

• seismic splitbam

• seismic importmm

• seismic list

seismic wf does not accept --branch directly, because it runs many steps at once. Instead, to branch one or more steps within the workflow, use --wf-branch STEP NAME, giving the name of the step followed by the branch name. Repeat the option to branch several steps in one run, e.g. --wf-branch filter strict --wf-branch cluster strict. The step name must be one of demult, align, idmut, filter, filterscan, cluster, clusterscan, or fold; any other name raises an error. See seismic wf.

See also

• Parallelize Tasks — parallelize across multiple inputs

• Define Regions — run the same step over different regions of a reference