This was a question that I thought was interesting enough to test ChatGPT with.

Surprisingly, it gave an answer along the lines of the parent comment.

However, it seems it didn't figure this out by itself but it says:

> It’s widely known in bioinformatics that “one-line Fasta” files compress much better with LZ-based algorithms, and this is discussed in forums, papers, and practical guides on storing genomic data.