Output files
SCARAB writes logs, feature matrices, clustering assignments, and FASTA products under -o. It does not provide the report/explorer interface of MetaPathways or ASPIRE.
OUTPUT/
├── SCARAB_log.txt
├── scarab_run.json
├── scarab_status.json
├── *.subcontigs.fasta
├── *.mbacov.tsv
├── *.coverage.scaled.tsv
├── feature/signature/parameter intermediates
└── <autoopt-method>/<selected-setting>/
├── *.tsv
├── denovo/
├── hdbscan/
├── ocsvm/
├── intersect/
└── xpgs/
The method and setting names are resolved by AutoOpt and printed in the log. Some intermediate names and optional products depend on which paths ran; this is a guide to the layout, not a promise that every directory will contain sequences.
Product |
Interpretation |
|---|---|
|
Raw MetaBAT depth-summary table used for abundance features. |
|
Scaled abundance matrix used by clustering. |
|
De novo assignments and unassigned/noise records. |
|
Trusted-anchor HDBSCAN assignments. |
|
One-class SVM recruitment assignments. |
|
Pairwise-agreement recruitment plus qualifying anchors; see the workflow set formula. |
|
Original contigs assigned to de novo bins. |
|
Recruited contig sets from the anchored approaches. |
|
Trusted reference plus recruits after BBTools deduplication. |
Recruited FASTAs and xPG FASTAs answer different questions: the first contains recruited metagenomic sequence; the latter also contains the trusted reference. The deduplication command uses a 97% minimum identity setting. Independently validate genome quality and interpret overlap between recruitment approaches.
The abundance stage removes SAM/BAM and selected tool-output intermediates after coverage generation. Retain final FASTAs, assignment tables, logs, input manifests, and the software revision for downstream review. Avoid manually pruning a partially completed output if you intend to reuse it.
For guided assemblies derived from these products, see reassembly.