Sharing a tool I built: biom-viewer, a native macOS viewer for .biom (Biological Observation Matrix) files.

The problem: .biom tables are sparse — mostly zeros. Every existing way to inspect one (biom convert to TSV, biom.load_table() + pandas, opening in Excel) densifies the entire matrix before you see anything, so a moderately sized table can blow past available RAM just to be looked at.

What it does: keeps the table sparse in memory and only densifies the observations/samples currently visible, so opening and scrolling stays fast regardless of file size. On top of that:

• Sample/observation metadata as first-class table views (not a separate export)
• Stackable filters on metadata fields (numeric range or category), shown as removable chips with live counts
• Inline summary stats per row/column (nonzero %, distribution histogram, min/max, top-values for categoricals), computed on the current filtered view
• Combined search across IDs, metadata field names, and metadata values

MIT licensed, source and macOS (Apple Silicon) builds here: https://github.com/yarintm/biom-viewer

Feedback and issues welcome — particularly interested to hear if it holds up on tables much larger than what I've tested against.