Just to update this, it is not recommended to use
fastq-dump for downloads. It is slow and prone to connection losses. Better use
prefetch together with Aspera, see here, to get the SRA files, and then use
fastq-dump to convert to fastq. Still, you can get most data directly from the European Nucleotide Archive in fastq format. Downloading from there is pretty simple and fast, see my tutorial on that: Fast download of FASTQ files and metadata from the European Nucleotide Archive (ENA) . If you have to download from NCBI, e.g. because data are restricted, go with
prefetch followed by parallel-fastq-dump, which is a wrapper for parallelizing
fastq-dump. After successfully converting a sra to fastq, both tools (
fastq-dump/parallel-fastq-dump) print a summary message that only shows up if no errors occurred, so I never felt the need to verify the fastq file after converting from sra, given that message was printed.