A lot of bioinformatics (my field) uses graph algorithms. DNA sequencing machines don't give you entire genomes, they give you pieces. Re-assembling the genome from those pieces is a prime graph algorithm target.

For example, Canu, one of the more popular current assemblers for PacBio and MinION data, uses a best overlap graph https://genome.cshlp.org/content/27/5/722.long

If you don’t mind the marketing aspect, there are some pretty good overviews in graphconnect (neo4j) conferences. https://m.youtube.com/watch?v=dW6JsFccdkM

There is a talk by two Facebook engineers [0] where they talk about using graph partitioning to better distribute caching of data.

[0]: https://m.youtube.com/watch?v=QHkhyY9atkE