Hi all,
I opened a PR for the first CS-FileSystem ATLAS metadata deliverable:
https://github.com/cyber-shuttle/CS-FileSystem/pull/1
This PR adds ATLAS metadata support to the FUSE filesystem. It loads ATLAS
entries from a local TSV file and exposes them under an /atlas directory,
with each protein entry represented as a directory containing a
metadata.json file.
Below is the example layout:
/tmp/atlas_mount/
atlas/
1r6w_A/
metadata.json
>From further feedback, I also refactored the ATLAS-specific filesystem
logic into atlas.rs behind a VirtualDataSource trait, so main.rs now acts
more like a generic FUSE router for top-level data sources. This should
make it easier to add other sources later, such as /s3 or /persona. The
testing done locally on macOS with macFUSE:
cargo test
cargo run --release -- data/2024_11_18_ATLAS_info.tsv /tmp/atlas_mount
Also, I've personally verified that the filesystem loads 1,938 ATLAS
entries and returns JSON metadata for entries like 1r6w_A. I would
appreciate any feedback on the structure, especially whether the
VirtualDataSource abstraction is heading in the right direction.
Thanks,
Jayanth Vennamreddy
