Step-by-step guide¶

This page is aligned to the code layout in SpaceTracer/steps so users can map tutorials directly to implementation files.

For exact input/parameter interpretation, open each step page under this section.

Code-aligned execution map¶

Step tutorial	Runtime step name	Purpose	Code step file
Step 0: Cluster Preparation	cluster	Load or compute spot/domain clusters.	step0_cluster.py
Step 1: BAM Processing	bam_processing	Build in-tissue filtered BAM for downstream analysis.	step1_bam_processing.py
Step 2: Pileup Candidate Sites	mpileup	Generate candidate loci and chunk metadata.	step2_mpileup.py
Step 3: Count and Prior Construction	umi_combine cell_num prior	Aggregate UMI-level count evidence, build cell-number support values, and construct prior information for candidate-site genotyping.	step3_UMI_combine.py step3_cell_number.py step3_get_prior.py
Step 4: Genotyping	genotyping	Infer genotype evidence at multiple levels.	step4_genotyping.py
Step 5: Feature Extraction and Phasing	spatial_feature mappability_feature read_feature RNA_feature phasing	Extract spatial, mappability, read-level, and RNA-level features, then refine candidate sites with phasing evidence.	step5_spatial_feature.py step5_mappability_feature.py step5_read_feature.py step5_RNA_level_feature.py step5_phasing.py
Step 6: Feature Merge and Filtration	merge_feature	Merge features and apply filtration tags/switches.	step6_merge_all_features.py
Step 7: Mutation Prediction	mutation_prediction	Run model inference and export VCF results.	step7_mutation_prediction.py

Post-processing outside SpaceTracer/steps:

• Post-step: Phylogeny (optional)

How to run¶

Recommended: one-command workflow¶

SpaceTracer run --config config.yaml

Advanced: inspect by code step order¶

Use these pages in code order:

• Step 0: Cluster Preparation
• Step 1: BAM Processing
• Step 2: Pileup Candidate Sites
• Step 3: Count and Prior Construction
• Step 4: Genotyping
• Step 5: Feature Extraction and Phasing
• Step 6: Feature Merge and Filtration
• Step 7: Mutation Prediction
• Post-step: Phylogeny (optional)

Important outputs by stage¶

• Code step 4: genotype evidence files (for example ind_geno_filter_file, spot_geno_file, cluster_vaf_file)
• Code step 6: merged/filtered feature tables (all_feature.txt and parquet mirror)
• Code step 7: predicted VCF outputs from mutation prediction
• Optional post-step phylogeny: lineage analysis from high-confidence loci

For practical rerun recipes (for example, rerun only RNA or spatial branches), see Single-Step Debug Cookbook.