Bioinformatics using the Cloud Storage with NCBI data

Table of Contents



Introduction

This guide is a reference on how to easily utilize all the different storage components of the NeCTAR into a bioinformatics framework so that you can work more efficiently. For the puposes of developing our framework will use the NCBI-BLAST+ as it is a popular application across many areas of bioinformatics. The BLAST data sets can be very large and that present challenges to store and access that data locally making it difficult to implement in the NeCTAR research clould.


Top of page


NeCTAR Research Cloud storage components review using the OpenStack Dashboard

The NeCTAR is a federated research cloud that uses OpenStack as the cloud operating system it controls the compute, storage and networking resources across the different zones that make up the research cloud. The OpenStack dashboard lets user provision resources for their instances through a web interface. The storage resources can be classified as attached and external disks each of which has different purpose that needs some basic explanation.


Attached storage

With your instance you have access to limited diskspace with the "root disk" partition "/" and "ephemeral disk" partition that is accessed via the /mnt directory of the default NeCTAR images. The size of the the "root disk" and "ephemeral disk" varies with the different hardware templates or [flavors] (https://support.nectar.org.au/support/solutions/articles/6000055380-resources-available-to-you) that are available when you are configuring your instance. So understanding your research data requirements is key so you can select the correct flavor for your instance. However, as with everything including the research cloud it does require a bit more insight of how a research might make the best use of these storage components in a bioinformatic workflow. Fortunately, NeCTAR is a the ideal resource to allow researchers to improve their technical knowledge of to how to use the research cloud for their computational work.


Top of page


What is the root disk

The root disk is an ephemeral disk that the operating system (OS) image is copied to when you launch a virtual machine. It is also where your home directory is located in you virtual machine. When you snapshot your instance it is what is copied to create your snapshot image.


Top of page


What is ephemeral disk

Ephemeral is a secondary ephemeral data disk. That is an empty disk and exists only for the life of the instance. When you terminate an instance all data in ephemeral is lost. With the NeCTAR default images that have ephemeral storage are mounted with the instance and located at /mnt but you will need to change the ownership of the /mnt directoryto be able to have read/write access to do this on a ubuntu instance use:


$ sudo chown ubuntu /mnt


on a fedora, centos, scientific linux instances


$ sudo chown ec2-user /mnt


and on a debian instance.


$ sudo chown debian /mnt


This ephemeral disk space is where you should run you analysis as your home directory is too small.


Top of page


External Storage - Volume and Object

When you apply for you NeCTAR allocation you are able to apply for object and volume storage to use for research project.



Assuming that your allocation request has either volume or object storage or both when you configure your volume or object storage you need to remember to do select a sensible availability zones to use with your data.


What is volume or block-storage

Persistent block-storage or volume storage can live outside of you instance. This means that you read and write data in the persistent volume storage and it can be moved around to different instances but can only be mounted to one instance at any given time. You can find more information about creating and attaching persistent volumes through the NeCTAR dashboard on the NeCTAR support page. Once it is attached to an instance the dashboard will show what the this:


note the volume names are hyperlinks that show the volume overview,



The Attachments show what device the that the volume is attached to in the is case

/dev/vdd

where /dev is the name of the device files and /vd* is the name of the virtuo block device.

The important commands to that you need to know to configure the persistent volume storage once you have attached to your instance. For most NeCTAR research cloud users an ext4 filesystem is more than sufficient and the different filesystems are beyond the scope of this document.


To to create an ext4 filesystem on the device or persistent volume use

$ sudo mkfs.ext4 /dev/vdd


WARNING: You only need to do this once as this reformats the device any data on the device will be lost!


Create a directory to mount the volume to it can be anything but should be meaningful and consistent for your work.

$ sudo mkdir /volume_data


To mount the persistent volume the command is:

$ sudo mount /dev/vdd /volume_data -t auto


Lastly you will need to change the ownership of the /volume_data directoryto be able to have read/write access to do this on a ubuntu instance use:

$ sudo chown ubuntu /volume_data


on a fedora, centos, scientific linux instances

$ sudo chown ec2-user /volume_data


and on a debian instance.

$ sudo chown debian /volume_data


To unmount the volume from your instance use these command umount NOTE: Do not umount the device if you are in that directory so cd to your home directory first otherwise you it will fail. 

$ cd
$ sudo umount /volume_data


Top of page


What is object storage?


A more useful definition can be found on the NeCTAR support page.

From the NeCTAR dashboard you can create containers in the Object Store.

click the create container button

Fill in the Container name and the click the create container button. You can then choose if you want to make the object container public or private.

The private objects are only available within your NeCTAR project. While object in public containers are exactly as the note states in the window when you created your container

It is available to anyone that knows the URL to that object including users who don't have access to NeCTAR instances.

You can upload object data from you local computer to the container using the dashboard just select the Upload Object button. To download an object to your local computer just select the Download button.

To download the object to your instance you just neeed to use wget

wget https://dashboard.rc.nectar.org.au/project/containers/blast/wgs.99.tar.gz


Top of page


File I/O best practices

Moving data to and from you instance instance depending on the size of the files your are try to move can be trival for small data but large data sets can become very difficult depending on many different things. Here are some things to keep in mind with data I/O and the cloud.

  • Compress your data files if they are larger than 10Mb with gzip or some other compression tool it should be installed in your instance.

  • If you have to move a directory create tar files.

  • Don't try to move data over a wireless network, plugin to an ethernet contection, preferably at your local institution they have larger bandwidth.

  • If you data move still time out or are getting throttled contact your the local institutes IT support to see if they can help.

  • Ohio-(Only handle it once) if you are uploading data to your instance try to move it directly to your instance, NeCTAR is on a fast networks as most nectar nodes are housed within University or HPC data centers.

With smaller data you can try any of the methods coverd on the NeCTAR support page it has excellent instructions. This will work for moving data to and from your ephemeral or volume storage attached to your instances.


Top of page


Using the OpenStack APIs for Managing NCBI Data

The National Center for Biotechnology Information (NCBI) is located in Bethesda, Marylanl and hosts the Basic Local Alignment Search Tool (BLAST) data and is available directly from the NCBI FTP site ftp://ftp.ncbi.nlm.nih.gov. It is necessary for researchers to be able to access these data sets easily as they are required for many different Bioinformatic tools specifically BLAST. However we will not present how to actually run BLAST but how to manage the NeCTAR reseach cloud storage resources to enable researchers to utilize the NCBI BLAST data sets effectively.


Install the Python client tools

First you will need to update your instance and install a set of packages that support the OpenStack APIs. These are the commands for a ubuntu instance.

$ sudo apt-get update

$ sudo apt-get install python-novaclient

$ sudo apt-get install python-keystoneclient

$ sudo apt-get install python-cinderclient

$ sudo apt-get install python-swiftclient


Top of page


Configuring your instance to use the Openstack API

This the basic procedure to configure within your instance to use the OpenStack APIs with the Python clients. The first step is to set or reset your your password from the OpenStack dashboard through your user account settings.


Select the Reset Password tab:



that takes you to the Password Reset Form where you reset your password



This is your new password it is case sensitive. We will use this in the OpenStack RC file.

From the OpenStack dashboard under the Compute -> Access & Security, select the Download OpenStack RC File that will automatically download a RC-file to your local machine.



The OpenStack RC File that you just downloaded will have the name Project_name-openrc.sh You will need to edit the following lines in the Project_name-openrc.sh file.

So that the new password is hardcoded into the Project_name-openrc.sh file.

You will need to the copy that file and your ssh key file either the ssh pem file or the ssh key file that is associated with your instance from your local machine to your instance. Using any of the method described in NeCTAR support documents. You should move your key file or pem file to your .ssh directory in your home directory on your instance.

On your instance you will need to source the Project_name-openrc.sh file to configure your environment to authenticate access to the OpenStack services.

$ source Project_name-openrc.sh


You will need to source the script before you can run any of the commands otherwise the environment is not set properly.


Gotchas: If you see this

you have forgotten to source the OpenStack RC file.

If you source your OpenStack RC file and see this:

You did not edit the OpenStack RC file!


Testing you API authentication

To test your access authentication you can use the command

$ nova list

this should show the list of your current instances that indicates your authentication access was successful!


Top of page


Downloading NCBI datasets directly to your instance

The data transferring program listed NeCTAR support documents work nicely from your local computer but can be difficult to implement in the cloud because of gui's or poor performance. What we need is a program to allow us to download from the NCBI ftp site directly to our instance's ephemeral or volume storage that works from the commandline.

NCBI has had good success with the lftp program available from the image packages repository has shown to work nicely from the UNIX/Linux commandline. 


To install it to your instance you can just use the package managment tools for ubuntu the command is:

$ sudo apt-get install lftp


You can then use the lftp command

$ lftp ftp.ncbi.nlm.nih.gov


Then you can use the same ftp commands to navigate between the local and remote systems and download with get and upload with put

This figure shows how simple this process is to get to the blast datasets



Now you can use get to download the data to your instance directly.


Using the swift API create a new container

To list all the containers in your project.

$ swift list

To create a new private container with the swift API from the command line

$ swift post container_name

To create a new public container the swift API use:

`$ swift post -r .r:* container_name``

To upload the data into the container.

$ swift upload 'container_name' 'data_file'

To see your new container

$ swift list

and the objects in your new container

$ swift list 'container_name'

Top of page

Summary

We have covered the difference between root, ephemeral, volume and object storage available through the NeCTAR research cloud and what programs are used to be able to store you data correctly so that you can make effective use of you virtual machine.

  • Root storage

Is for the Operating system and you home directory. You should install all applications there and your home directory is should not store any data but small scripts and files.
  • Ephemeral storage

Is your workspace so all of you analysis should be done there and you can store your local data there. Remember to get the right size hardware flavor. It needs to have enough disk space for analysis to run and store any associated data. It is your responsibilty to maintain and will require regular housekeeping. Please delete files that are not needed and move important results back to either your local machine or to volume/object storage to share them with your group or the world.
  • Volume storage 

Can be used for storing results, working data, or as a working directory if needed. If you have multiple researcher who have access to your project then you can share data via the volume storage. It is also possible to attach more that one volume storage instance to your compute instance as well so you can one volume for input data and have a second volume for results.
  • Object storage 

Should be used for static datasets either reference ones like the NCBI-BLAST data sets or analysis results you wish to share with collaborators. Static data can be downloaded easily to your volume or ephemeral storage as part of you workflow or pipeline. With an original copy in object storage you can delete it from either volume or ephemeral storage then with out having to get a new copy to your instance.

NOTE: There are some unresolved issues with uploading large data sets (>5GB) in to object storage. Understanding how to split/merge files that are larger than 5GB and being able to manage them easily..

Top of page