Hi James,
The approach you describe here looks correct, and that's how I typically combine samples when I need to quantify them together. The only extra trick I'd add is that, if you're using enough quantification threads, Salmon can actually process the reads faster than they can be decompressed. Thus,
you might see a speedup if you replace `zcat` with `pigz -d -c -n <num_decompression_processes>` — this will do the same thing, but in parallel.
Best,
Rob