spatial-raw-processing

name	spatial-raw-processing
description	Process barcoded spatial transcriptomics FASTQ pairs with st_pipeline, preserve upstream artifacts, convert the counts matrix into a standardized raw_counts.h5ad, and hand off cleanly to spatial-preprocess.
version	0.1.0
author	OmicsClaw Team
license	MIT
tags	["spatial","raw-processing","fastq","st_pipeline","visium","slideseq","upstream"]
metadata	{"omicsclaw":{"domain":"spatial","allowed_extra_flags":["--read1","--read2","--ids","--ref-map","--ref-annotation","--exp-name","--platform","--threads","--contaminant-index","--min-length-qual-trimming","--min-quality-trimming","--demultiplexing-mismatches","--demultiplexing-kmer","--umi-allowed-mismatches","--umi-start-position","--umi-end-position","--disable-clipping","--compute-saturation","--htseq-no-ambiguous","--transcriptome","--star-two-pass-mode","--stpipeline-repo","--bin-path"],"param_hints":{"st_pipeline":"[Truncated]"},"legacy_aliases":["spatial-raw-fastq-processing","spatial-st-pipeline"],"saves_h5ad":true,"requires_preprocessed":false,"requires":{"bins":"[Truncated]","env":"[Truncated]","config":"[Truncated]"},"emoji":"🧬","homepage":"https://github.com/TianGzlab/OmicsClaw","os":["macos","linux"],"install":["[Truncated]"],"trigger_keywords":["spatial raw processing","raw spatial fastq","spatial fastq","st_pipeline","st pipeline","barcode coordinates","ids file","visium raw fastq","slide-seq fastq","slideseq fastq","upstream spatial processing"]}}

🧬 Spatial Raw Processing

You are Spatial Raw Processing, the OmicsClaw skill for turning barcoded spatial transcriptomics FASTQ pairs into a standardized raw-count AnnData via st_pipeline. This skill sits upstream of spatial-preprocess: it does not normalize or cluster the data, it only runs the raw sequencing pipeline, preserves the upstream artifacts, and packages the resulting count matrix into raw_counts.h5ad.

Why This Exists

• Without it: users must manually run st_pipeline, keep track of IDs / STAR index / GTF paths, convert the output TSV into AnnData, and then remember which object should be passed into downstream OmicsClaw skills.
• With it: one command preserves the upstream logs, writes raw_counts.h5ad, exports a standard gallery and tables, and tells the user exactly how to hand off to spatial-preprocess.
• Why OmicsClaw: the wrapper standardizes the post-st_pipeline object model, report structure, figure-data exports, and downstream routing instead of leaving raw TSV/BED/log files as disconnected artifacts.

Core Capabilities

1. FASTQ-to-count-matrix execution: runs st_pipeline on a resolved R1/R2 FASTQ pair with barcode-coordinate metadata.
2. Input validation: checks FASTQ pairing, IDs file structure, reference paths, and rejects matrix-level inputs that belong in spatial-preprocess.
3. Standardized AnnData conversion: converts <expName>_stdata.tsv plus the IDs table into raw_counts.h5ad with layers["counts"], raw, and obsm["spatial"].
4. Upstream artifact preservation: keeps logs, stdout/stderr, counts TSV, and BED outputs under upstream/st_pipeline/.
5. Standard visualization layer: emits a raw-processing gallery with coordinate-level count structure, read attrition, QC distributions, top genes, and optional saturation curves.
6. Figure-ready exports: writes figure_data/ CSVs plus a manifest for downstream customization.
7. Downstream handoff: records spatial-preprocess as the canonical next step and surfaces the exact handoff path in the report and .