11 MB
7 files
Updated 2 days ago
Name
Size
.gitattributes2.5 kB
xet
CHANGELOG.md4.87 kB
xet
LICENSE736 Bytes
xet
README.md2.68 kB
xet
data.zip11 MB
xet
problems.csv3.14 kB
xet
problems.parquet7.23 kB
xet
README.md

BioMysteryBench (preview)

A 5-problem preview of BioMysteryBench, a bioinformatics research benchmark created by Anthropic. Each problem provides anonymized biological data files and asks a question that requires real analysis to answer — the source dataset cannot be looked up.

v11 (2026-07): preview refreshed — hb022 and hb053 were removed from the benchmark; hb024 and hb035 replace them here. See CHANGELOG.md.

Contents

  • problems.csv / problems.parquet — one row per problem (id, question, answer_rubric, allowed_domains, human_solvable)
  • data.zip<id>/… data files for all 5 problems; extract into the working directory before solving

Rules

Looking up GEO/SRA/ENA/BioProject accession IDs to identify the original dataset, source publication, or study metadata is not allowed. Standard bioinformatics database use (gene ID lookup, sequence annotation, reference genome download, BLAST) is allowed.

Problems in this preview

id human_solvable question
hb020 yes What organism does this crystal structure belong to? Provide the answer using binomial nomenclature …
hb002 yes What bacteria is found in this sequenced dataset? Provide the bacteria's scientific name as the answ…
recqgsfxqqodhjens yes Identify the transcription factor whose binding sites are represented by the provided peaks data. Pr…
hb024 no This dataset involves microbial communities from tissues of non-human species. How many tissue group…
hb035 no What is the name of the most common microRNA associated with the genes given in this data? (Give the…

The full 90-problem set is at Anthropic/BioMysteryBench-full (gated, evaluation-only).

Total size
11 MB
Files
7
Last updated
Aug 10
Pre-warmed CDN
US EU US EU

Contributors