5.5 Per-Sample Batch Mode

In per-sample batch mode:

• The system scans the BioLine directory and detects samples from subdirectories, paired naming, or single-file patterns.
• Shared files, such as the host reference genome and metadata, are identified separately rather than uploaded repeatedly for every sample.
• Each sample receives its own BioLine pipelineId and state.
• Cloud output names include the Job and sample identifiers, for example open-<job-id>-<sample-id>.
• Downloaded outputs are placed under stage/<category>/<sample-id>/ to avoid collisions.
• Overall batch state is stored in .batch_state.json.