templates/
├── slurm/      # SLURM job scheduler templates
├── pbs/        # PBS/Torque job scheduler templates
└── local/      # Local machine templates (with GNU parallel support)

Template Naming Convention

Templates follow a consistent naming pattern: NN_step_name[_variant].ext

NN: Step number (e.g., 02 for compleasm, 08a for partition search)
step_name: Descriptive name of the pipeline step
_variant: Optional variant (e.g., _first, _parallel, _serial)
.ext: File extension (.job for schedulers, .sh for local scripts)

Available Templates

Step 2: Ortholog Identification (compleasm)

SLURM:

02_compleasm_first.job - Process first genome to download lineage database
02_compleasm_parallel.job - Array job for remaining genomes

PBS:

02_compleasm_first.job - Process first genome to download lineage database
02_compleasm_parallel.job - Array job for remaining genomes

Local:

02_compleasm_first.sh - Process first genome to download lineage database
02_compleasm_parallel.sh - GNU parallel for remaining genomes

Step 8A: Partition Model Selection

SLURM:

08a_partition_search.job - IQ-TREE partition model search with TESTMERGEONLY

PBS:

08a_partition_search.job - IQ-TREE partition model search with TESTMERGEONLY

Local:

08a_partition_search.sh - IQ-TREE partition model search with TESTMERGEONLY

Step 8C: Individual Gene Trees

SLURM:

08c_gene_trees_array.job - Array job for parallel gene tree estimation

PBS:

08c_gene_trees_array.job - Array job for parallel gene tree estimation

Local:

08c_gene_trees_parallel.sh - GNU parallel for gene tree estimation
08c_gene_trees_serial.sh - Serial processing (for debugging/limited resources)

Placeholders

Templates contain placeholders that must be replaced with user-specific values:

Placeholder	Description	Example
`TOTAL_THREADS`	Total CPU cores available	`64`
`THREADS_PER_JOB`	Threads per concurrent job	`16`
`NUM_GENOMES`	Number of genomes in analysis	`20`
`NUM_LOCI`	Number of loci/alignments	`2795`
`LINEAGE`	BUSCO lineage dataset	`insecta_odb10`
`MODEL_SET`	Comma-separated substitution models	`LG,WAG,JTT,Q.pfam`

Usage

For Claude (LLM)

When a user requests scripts for a specific computing environment:

Read the appropriate template using the Read tool
Replace placeholders with user-specified values
Present the customized script to the user
Provide setup instructions (e.g., how many genomes, how to calculate thread allocation)

Example:

# Read template
template = Read("templates/slurm/02_compleasm_first.job")

# Replace placeholders
script = template.replace("TOTAL_THREADS", "64")
script = script.replace("LINEAGE", "insecta_odb10")

# Present to user
print(script)

For Users

Templates are not meant to be used directly. Instead:

Follow the workflow in SKILL.md
Answer Claude's questions about your setup
Claude will fetch the appropriate template and customize it for you
Copy the customized script Claude provides

Benefits of This Structure

Reduced token usage: Claude only reads templates when needed
Easier maintenance: Update one template file instead of multiple locations in SKILL.md
Consistency: All users get the same base template structure
Clarity: Separate files are easier to review than inline code
Extensibility: Easy to add new templates for additional tools or variants

Adding New Templates

When adding new templates:

Follow naming convention: NN_descriptive_name[_variant].ext
Include clear comments: Explain what the script does
Use consistent placeholders: Match existing placeholder names
Test thoroughly: Ensure placeholders are complete and correct
Update this README: Add the new template to the "Available Templates" section
Update SKILL.md: Reference the new template in the appropriate workflow step