sThis is for the one-time installation of the ctyper database on Azure Cyclecloud clusters.

1. Preliminaries; cyclecloud setup.

The Azure CycleCloud Workspace for Slurm should be created as a resource. This will create the /shared NFS directory visible to all nodes. To be safe, plan for 2-3 times the total storage space of all the assemblies (they will need to be copied to the cyclecloud).

The image should be Ubuntu 22.04, slurm 25.05.2 (current defaults with Cyclecloud)

The resources that we have used are:

Scheduler VM Type Standard_D8as_v4 Login node VM Type Standard_D8as_v4

HTC VM Type Standard_E64-16s_v6, Standard_D8s_v6 The maximum number of nodes should scale with what you are comfortable running

The pipeline does not use HPC/GPU nodes, the maximum number of nodes for these should be set to 0.

The Extend_ori directory should be copied to /shared

2. Environment setup.

2.1 g++

Most of the environment is set up using conda, however some binaries are built from source, which requires g++ to be installed.

sudo apt update
sudo apt install -y build-essential
sudo apt install -y git

2.2 conda

You will want to install conda in a location that is accessible on NFS.

wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh
chmod +x ./Miniconda3-latest-Linux-x86_64.sh
./Miniconda3-latest-Linux-x86_64.sh

Follow the instructions, and ensure that the root directory is under /share

2.3 Installing the database construction scripts

First, change to the Extend_ori directory: cd /share/Extend_ori

Next, clone this repository, and run the setup script:

cd extend_ctyper_database
source setup_env.sh
cp -r scripts ../
cp Snakefile ..```

Your environment to run database construction is complete. Within the directory /share/Extend_ori, you should have a folder scripts, and a file Snakefile.

3. Configuring database construction build.

The instructions on configuring files for building the database are available here: https://github.com/Walfred-MA/PATs/

The only difference is that the config.json file that needs to be created does not need to specify a path for scripts. The following can be used for a config.json file to appropriately run Snakemake.

{
  "slurm": " ",
  "QueryPath": "query_pathes_withrefs.txt",
  "TargetFolder": "groups",
  "TempFolder": "snaketemp",
  "genelist": "./genes.bed",
  "ReferencePrefix": "NC_0609",
  "NumPartitions": 1
}

4. Set up files.

Create a directory /share/Extend_ori/Assemblies, and copy all of the repeat masked assemblies that will be added to the database to this folder. The assemblies should have one FASTA file per haplotype as produced by hifiasm. Assemblies produced with older assembly algorithms that produce primary and secondary contigs are not compatible with ctyper.

You will create a file query_pathes_withrefs.txt (spelling needs to be exact, and to match what is in the .json file for QueryPath) that has two columns, one the name of the sample and haplotype (e.g. sample001_h1), and the other the path to the assembly (Assembly/sample_00_h1.fasta). You can follow the example here: https://github.com/Walfred-MA/PATs/?tab=readme-ov-file#b-create-the-assembliesreference-path-list-tsv

5. Run database construction

snakemake -k --cluster "sbatch --partition htc --time=500:00:00 {resources.slurm_extra}" --default-resources "mem_mb=3000" --jobs 500 --rerun-incomplete --notemp --latency-wait 100 --resources mem_gb=1000

6. Merge database files.

Once the snakemake run is done, the results need to be merged. You can follow step F here: https://github.com/Walfred-MA/PATs/?tab=readme-ov-file#f-find-merge-and-index-compiled-k-mer-matrices