Input/output: work with compressed FASTQ files
All or most rules using reads work with decompressed FASTQ files. Especially for deeply sequenced samples, this increases the runtime, and requires more space (even if only temporary for all but the final FASTQ files).
Nowadays, most tools included in
cutadapt etc.) should be able to use compressed files.
Decompression should be done only when necessary to minimize the overhead.