This repository collects the RNA-Seq, Variant calling, and Single Cell benchmarks that are scattered amongst several other repositories including, but not limited to:
- Resource Prediction Experiments
- Biology of Genomes
- Experiments
Each of the rnaseq
, variant
, and single-cell
directories contain the benchmark, experiment, and TPV configuration files required to run the benchmark experiments.
costs.sh
Environment variable definitions used when creating a GKE cluster. See Variables below for a description of the variables used.
cascade-lake.sh
Variable definitions used for the clusters using Intel Cascade Lake cpus.
ice-lake.sh
Variable definitions used for the clusters using Intel Ice Lake cpus.
anvil
is a Bash script used to manage the GKE clusters used to run the experiments. Run anvil --help
for full usage information.
Google supports specifying a --min-cpu-platform
, but that is a minimum and we may get a later CPU family if the minium requested family is not available. Therefore we need to ensure the CPUs used in the cluster are actually the type requested.
# SSH into the node
gcloud compute ssh --zone <ZONE> <instance name>
# Install GCC
sudo apt update
sudo apt install build-essential
# Get the CPU family name
gcc -march=native -Q --help=target | grep -march
anvil --settings settings.sh cluster disks galaxy
anvil --settings settings.sh cleanup delete-disks
If you do not delete the persistent disks they can be re-attached to a new cluster as long as the same cluster --prefix
is used. The easiest way to achieve this is to use a settings
file that contains definitions for the common variable definitions.
anvil --settings settings.sh cluster disks galaxy
anvil --settings settings.sh cleanup
anvil --settings settings.sh cluster galaxy # Will re-use the persistent disks created in step 1
The following environment variables can be set that will be used by the anvil
Bash script.
MACHINE
Machine type for cluster nodes, e.g. n2-standard-32
.
PREFIX
A prefix string used to when creating resources.
NAMESPACE
Kubernetes/Helm namespace used when installing Galaxy. Defaults to galaxy
.
IMAGE
Docker image used to install Galaxy. Defaults to quay.io/galaxyproject/galaxy-anvil
.
TAG
Image tag of the Docker image used to install Galaxy. Defaults to 23.1
.
CHART
Helm chart used to install Galaxy. Defaults to anvil/galaxykubeman
.
DISK
Size of the persistent disk created. In GB. Defaults to 768
.