SRA Toolkit overview
SRA (Sequence Read Archive) is an NCBI-defined interchange format for NGS data. The idea is that before submitting your data to NCBI, you convert whatever format it is in (fastq, bam, etc.) to SRA format using one of the "load" tools. Then, the data can be downloaded from NCBI by anyone and extracted in one of a number of different formats as desired (ABI csfasta/qual, fastq).
While this sounds like a great idea (someone else taking care of format interchange issues for you!), the toolkit is no longer being actively developed except for bug fixes. However there is a lot of interesting data out there that's only available as SRAs so it is worthwhile knowing how to use it.
For example, you have aligned a ChIP-seq dataset to hg19 and have a .bam file. You want to upload the data to NCBI. You use the bam-load tool:
The raw reads can be then be extracted to fastq using fastq-dump:
Looks deceptively simple but you can run into problems. For one thing, SRA toolkit versions change often and are not always compatible. So if you get any weird errors, check for a newer (or sometimes older) toolkit version. The SRA Toolkit documentation, such that it is, is located at the NCBI website.
Submissions for a publication generally have the form SRPnnnn, with all the data under an accession SRAnnn (the n's have no relation to one another). Data is organized by experiment (SRXnnnn) and sequencing run (SRRnnnn).
The SRA search home page is where to start looking.
Find and download RNAseq data from run SRR390925, of experiment SRX112044, publication SRP009873. Copy the file to your home directory on Lonestar at TACC then extract the data in fastq format.
SRA search home page http://www.ncbi.nlm.nih.gov/sra
module spider sratoolkit