Instructor Notes

General Setup

Before the workshop:

  1. Verify all training accounts (trainXX) can access both Negishi and Gilbreth clusters
  2. Confirm the shared data is available at /depot/workshop/data/annotation_workshop/
  3. Ensure biocontainers module loads correctly and all tool modules are accessible
  4. Pre-stage the GeneMark license key for learners or provide download instructions
  5. Test that GPU allocation works on Gilbreth for the Helixer episode

Timing Guide

Episode Teaching Exercises Notes
Introduction 20 min 0 min Conceptual, no hands-on
Annotation Strategies 20 min 5 min Conceptual with exercises
Annotation Setup 15 min 5 min Walk through only; steps are pre-computed
Gene Annotation with BRAKER 40 min 10 min Submit jobs early; review results while waiting
Gene Annotation with Helixer 20 min 5 min Requires Gilbreth cluster; switch clusters during break
Gene Annotation with EASEL 25 min 5 min Optional/self-guided if time is short
Functional Annotation with EnTAP 20 min 5 min Databases are pre-staged
Annotation Assessment 30 min 10 min Compare all methods together

Common Learner Issues

Cluster Access

  • Learners may forget to switch to Gilbreth for the Helixer episode. Remind them during the break before that session.
  • SSH key setup can be tricky on Windows. Have learners test access during the Arrival & Setup period.

Module Loading

  • Always use ml --force purge before loading new modules. Without --force, sticky modules may cause conflicts.
  • If biocontainers fails to load, check that the user’s .bashrc does not have conflicting module commands.

BRAKER3

  • The GeneMark license key must be in ~/.gm_key. If learners get a “license key not found” error, have them re-copy the key.
  • AUGUSTUS config must be writable. Ensure the copy_augustus_config step completed successfully.
  • Jobs take 7-170 minutes depending on the case. Submit all 5 cases at the start and check results later.

Helixer

  • GPU availability on Gilbreth may be limited. Use --account=standby for opportunistic scheduling.
  • Model download to ~/.local/share/Helixer can be slow. Consider pre-staging models.

EASEL

  • Nextflow version matters. The custom installation step avoids conflicts with the system module.
  • The rcac.config must be correctly placed and referenced in nextflow.config.
  • If the pipeline fails, check .nextflow.log and the work/ directory for process-level logs.

EnTAP

  • Databases are large and pre-staged at /depot/itap/datasets/entap_db/. Learners should not re-download them.
  • The entap_config.ini paths must point to the correct database locations.

Assessment

  • BUSCO lineage dataset for Arabidopsis is brassicales_odb10. Double-check learners use the correct lineage.
  • The OMArk database (LUCA.h5) is pre-staged at /depot/itap/datasets/omark/.

Challenge Exercise Answers

Answers to all challenge exercises are included in the :::::: solution blocks within each episode. Review these before the workshop to anticipate follow-up questions.