Hello all,
For the variant calling purpose using gatk in human data, which would be better, a server or aws cloud? What will be specifications and configurations needed for both? Your suggestions would be greatly appreciated. Thank you.
I think that is more a matter of convenience and legality. It mostly depends on who pays for it and what you prefer. Both solutions should be equally easy to get GATK running via conda or a container. As you mention human data, in many countries specific data privacy laws cover this case. In Europe, transferring and processing human (patient) genetic data on AWS would almost certainly be illegal due to GDPR. I don't know what the specific laws and regulations are in India when it comes to handling and safeguarding human genetic data but you better check beforehand.
I assume you mean "server" as in local hardware that you own/control since "cloud" is a just bunch of servers on internet that you rent (part or whole) as needed. Choice between the two options may depend on local security restrictions. If these are clinical samples then "cloud" may not be an option if local policies prohibit such use.
Exact specs will depend on your budget but a minimum config would be at least necessary to run one sample at a time with no consideration of how long it takes. 32 GB of RAM with at least a TB (multiple needed depended on number of samples) of storage with a multi-core CPU (perhaps 16 cores min). Practical upper end limit would basically be dictated your budget. In theory you could rent N virtual machines (N = your sample number) and be done in a day.
Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
version 2.3.6