This repository contains course materials for a workshop on structural genome annotation with BRAKER, GALBA, and TSEBRA. The course is part of the Workshop on Genomics in Cesky Krumlov, Czech Republic (https://evomics.org/2024-workshop-on-genomics-cesky-krumlov/).

Authors: Katharina Hoff & Natalia Nenasheva

Contact: [email protected]

(Slide number may be slightly off if you look at the slides on GitHub due to remove slides, they correspond to what's shown on the projector!)

Slide 35: Tool name collection https://www.menti.com/7zgomub8sx

Slide 38: Tool categorization https://padlet.com/katharinahoff1/tools-for-structural-annotation-of-protein-coding-genes-c8desilwro11qp0h

Slide 78: Poll on understanding annotation methods

• theory: repeat library generation and repeat masking with RepeatModeler2/RepeatMasker
• theory: short read RNA-Seq to genome alignment with Hisat2
• theory: sorting an RNA-Seq alignment file with Samtools
• practice: application of BRAKER3 (structural genome annotation with RNA-Seq alignments and a large protein data base)
• practice: application of BRAKER1 (structural genome annotation with short read RNA-Seq alignments)
• practice: application of BRAKER2 (structural genome annotation with protein database)
• practice: application of GALBA (structural genome annotation with proteins of a closely related species, suitable for e.g. vertebrate genomes)
• practice: merging BRAKER1 and BRAKER2 (or GALBA) gene sets with TSEBRA
• practice: BUSCO assessment of predicted gene set
• practice: preparing an assembly hub for the UCSC Genome Browser with MakeHub
• for advanced learners: annotate a chromsome of Basesia duncati

1. Open a terminal and ssh into your instance

2. In your terminal, in your home (~), make a new directory for the git clone

mkdir your_name_GA # replace your_name_GA by your actual name, e.g. katharina_hoff_GA if your name is katharina hoff
cd your_name_GA
git clone https://github.com/KatharinaHoff/GenomeAnnotation_Workshop.git

This will create a folder called GenomeAnnotation_Workshop in your home directory. This folder contains the JupyterNotebook for this course (GenomeAnnotation.ipynb) and all the necessary data ☺️

The organizers of the Cesky Krumlov Workshop on Genomics have already compiled a singularity file called genome_annotation.sif for you. You can find this file at /home/genomics/workshop_materials/genome_annotation.

With the cloned data and the singularity file (genome_annotation.sif), you can run the image for JupyterNotebook display in the terminal as follows:

3. Execute singularity:

singularity exec --cleanenv --bind /home/genomics/workshop_materials/genome_annotation:/home/genomics/workshop_materials/genome_annotation --bind ${PWD}:${PWD} --bind $PWD:/home/jovyan /home/genomics/workshop_materials/genome_annotation/genome_annotation.sif jupyter notebook --no-browser --ip=0.0.0.0 --port=8899

This will display 3 links in your terminal. The links will look something like this:

http://127.0.0.1:8899/?token=1d5886ad8013bbcaeba6ccaef3dc815e91e17caa696ab596

Change the IP (127.0.0.1) to your instance address

DO NOT CLOSE YOUR TERMINAL! 🖥 It's essential that you keep it open. Click on the folder to access the workshop content. Double click to open the GenomeAnnotation.ipynb. Welcome to the starting point of this lab 🤓

You can run code by clicking "Run" at the top of the screen

Alternatively, you can click inside any box which starts with

%%script bash

and type Shift + Enter

You can stop a code block running by pressing "Stop" (the black square)

When a code block (the kernel) is running, it will look like this:

In [*]

When a code block has finished running:

In [num]

where num is an index of the job

When a code block has not yet been run:

In[ ]

To create a new code block, you click on "Insert"

If you want to obtain the same image for using it after the course, you can do so as follows (with singularity-ce version 3.11.2, available from https://github.com/sylabs/singularity, find their installation instructions at https://github.com/sylabs/singularity/blob/main/INSTALL.md, make sure you are not using an older version of singularity, as this may cause problems):

# execute from your user home directory, should not be a group drive
singularity exec --cleanenv --bind /home/genomics/workshop_materials/genome_annotation:/home/genomics/workshop_materials/genome_annotation --bind ${PWD}:${PWD} --bind $PWD:/home/jovyan /home/genomics/workshop_materials/genome_annotation/genome_annotation.sif jupyter notebook --no-browser --ip=127.0.0.1

The local directory /home/genomics/workshop_materials may only be available during the course on site at Cesky Krumlov's AWS instance. If you want to use the image after the course, you may want to remove this directory from the command above (explicitely, remove: --bind /home/genomics/workshop_materials/genome_annotation:/home/genomics/workshop_materials/genome_annotation). Also, you of course need to specify the true location of the image, modify the command if it does not reside at /home/genomics/workshop_materials/genome_annotation/genome_annotation.sif.

It is vital that you mount /home/jovyan to a writable location. Otherwise, you will not be able to save your work. The command above will mount the current working directory to /home/jovyan. If you want to mount a different directory, replace ${PWD} with the path to the directory you want to mount (this corresponds specifically to this part of the command: --bind ${PWD}:/home/jovyan).

The flag --cleanenv makes sure that other environment variables/tools (e.g. Perl dependencies) installed on the host do not interfere with the image.

This will display a link in your terminal that you may post into your web browser. The link will look something like this:

http://127.0.0.1:8888/?token=4aff4819888e4afd61a63b3015f8a1f816deea84efe2cd3f

If you want to execute the JupyterNotebooks, you will need data. At the Cesky Krumlov Workshop, these datasets have already been prepared for you at /home/genomics/workshop_materials/genome_annotation. If you want to use the JupyterNotebook after the course, you will need to download the data sets to your local device. Simply execute obtain_data.sh in your terminal (e.g. bash obtain_data.sh). This will require 28 GB of free space!

The exact same container that is used for rendering our teaching materials with JupyterLab can be used on any HPC (that has singularity support) with a SLURM scheduler. In this case, you will not execute JupyterLab, but submit a task with SLURM for computation. Example for calling BRAKER3 with SLURM:

#!/bin/bash                                  
#SBATCH -o braker.%j.%N.out
#SBATCH -e braker.%j.%N.err
#SBATCH --get-user-env
#SBATCH --time=72:00:00
#SBATCH -N 1 # number of nodes, BRAKER does not scale across multiple nodes
#SBATCH -n 48 # number of threads on that node, BRAKER does not scale well to hundreds of threads, we often execute on 8-48 threads

module load singularity

(singularity exec -B $PWD:$PWD genome_annotation.sif braker.pl --genome=genome.fasta.masked --prot_seq=proteins.fa --bam=rnaseq.bam --threads 48 ) &> braker3.log

Any tasks from our JupyterNotebook cells can be implemented in such scripts.

Simply submit the job with sbatch:

sbatch braker3.sh

Stefan Kemnitz from The University Compute Center at University of Greifswald (https://rz.uni-greifswald.de/dienste/allgemein/sonstiges/high-performance-computing/) kindly assisted in building docker containers for genome annotation with methods developed at University of Greifswald.

Josie Paris from Università Politecnica delle Marche provided very helpful instructions on how to use the cloud computing infrastructure during the Cesky Krumlov workshop.