Set up and run FastaLake with an assistant

An agent or LLM can help install FastaLake, prepare inputs, run a study and explain the results. Give it the checked-out documentation and your actual paths so it can work from the same commands you would use.

Ready-to-copy task files

Open the repository in your agent, ask it to read AGENTS.md, then give it one of these files with your paths filled in. The same text works in a chat: run the proposed commands yourself and return the output for checking.

Task	Prompt file	Example input
Install and inspect the demo	install-and-demo.md	Bundled demo
Run your own acquisitions	first-study.md	Working TSV
Add matched DNA/RNA	molecular-inputs.md	Molecular example
Size and submit a cluster pilot	cluster-pilot.md	Site template
Diagnose a failure	troubleshoot.md	Your failed run and stage logs
Understand a completed study	read-results.md	Your matrix, peptide evidence and QC

Each file contains an input checklist, the relevant guides, the requested work and the checks needed to call it finished. Browse the whole task folder.

flowchart TD
    A["Choose a task and fill in your paths"] --> B["Assistant reads AGENTS.md and the relevant guide"]
    B --> C["Check the environment and inputs"]
    C --> D["Run the demo or a small pilot"]
    D --> E["Inspect outputs and explain the result"]
    E --> F["Continue with the study using measured settings"]

What to tell the assistant

Supply the computer or cluster, repository path, input locations, new output directory and resource limits. For a real study, include the acquisition manifest and any specimen mapping. Ask it to explain missing inputs before guessing scientific settings.

A coding agent with terminal access can inspect and run the workflow. A chat assistant can guide you through commands and review the output you return. Establish which machine and files it can actually reach.

What a completed task should give you

• The revision, environment and exact command.
• Result paths, including the study matrix, QC and peptide evidence.
• Checks performed, skipped steps and unresolved questions.
• For cluster work, the job status and measured memory, time and disk.

The task files ask for these checks explicitly. A generated plot or a submitted job alone does not show that the workflow finished.