Show HN: tgv: Genome viewer for you and your agents

0.4.0.mp4

tgv (Terminal Genome Viewer) is blazing-fast. Your SSH session is no longer a black box.

- Navigate genomes with vim-style commands (but the mouse works too).

- Rich file format support: BAM, VCF, BCF, BED, bigBed; object storage (s3); and any UCSC reference genome.

- tgv GUI coming soon.

tgv organizes messy omics data to a performant data engine that's fully exposed to agents through MCP. A multi-omics analysis takes a few lines of SQL queries.

- No more glue scripts chaining samtools,bcftools, andawk.

- No more off-by-one bugs from mixing 0-based and 1-based tools.

- No more wasted tokens.

Note

tgv is in early development. Please report bugs and we will fix them asap.

- cargo: cargo install tgv --locked

- brew: brew tap zeqianli/tgv && brew install tgv

- bioconda: conda install bioconda::tgv

- Pre-built binaries: GitHub Releases

Install tgv MCP:

- Codex: codex mcp add tgv -- tgv mcp

- Claude: claude mcp add tgv -- tgv mcp

- Others: ask your agent

# Browse the hg38 human genome (internet needed)

tgv- :q: Quit

- h/j/k/l: Left / down / up / right.- H/J/K/Lfor faster navigation

- W/B/w/b: Next gene / previous gene / next exon / previous exon

- z/o: Zoom in / out

- /_gene_/- /_chr_:_position_: Go to a gene (e.g.- /TP53) or a position (e.g.- /1:2345)

- _number_+- _movement_: Repeat movements (e.g.- 20B: back 20 genes)

- :ls: Switch chromosomes

- :e _file_: Open more files, or drag them into the terminal

- Mouse: click, scroll, drag and hover.

If you use a reference genome frequently, downloading a local cache is highly recommended. This makes TGV much faster.

# The cache is in ~/.tgv by default.

tgv download hg38Browse alignments:

# View file aligned to the hg38 human reference genome

tgv file1.sorted.bam s3://my-bucket/file2.sorted.bam variants.vcf intervals.bed

# BAM file with no reference genome

tgv non_human.bam -r 1:123 --no-reference