From Python and R

Get a key

Open Account, then API keys, and click New key. Choose Read, or Read and add data when your code uploads. A key works as you: it opens what you can open on the site and nothing more. It is shown once; set it as BASEQ_API_KEY where your code runs.

Install

pip install "baseqdb[pandas]"                   # Python 3.10 or later

# R 4.2 or later, with no Python needed
install.packages("baseqdb",
                 repos = c("https://baseq.bio/r", "https://cloud.r-project.org"))

A dataset

import baseqdb as bq
bq.login()                                      # reads BASEQ_API_KEY

bq.Dataset.filter(kind="bulk").to_dataframe()   # yours, newest versions
ds = bq.Dataset.get("kTIqMTTm89P2gbNU0000")     # one version, by its uid
ds.versions                                     # every version you see
folder = ds.cache()                             # its files, downloaded once
sheet = ds.sample_sheet                         # a DataFrame

In R the same names, with $:

library(baseqdb)
bq_login()
ds <- Dataset$get("kTIqMTTm89P2gbNU0000")
folder <- ds$cache()
sheet <- ds$sample_sheet                        # a data.frame

Analyses and what they found

bq.Analysis.filter(kind="bulk").to_dataframe()  # every analysis you can open
analysis = ds.analyses[0]                       # or those of one dataset
analysis.steps.to_dataframe()                   # each step, its settings, when it ran
analysis.findings.to_dataframe()                # what each step found
table = analysis.tables["dge_condition_dexamethasone_vs_untreated.csv"].load()

Where a result came from

record = bq.Record.get("Z3HMY4V1wC6goA380000")  # a file a step wrote
record.run                                      # its step, settings, environment
record.run.inputs                               # the records it read

Add a version

With a Read and add data key. A dataset you add goes through the same checks as an upload; with the name of one you have, it is its next version, and it starts what When a new version is added says.

new = bq.Dataset(["counts_filtered.tsv"], name="Airway demo (GSE52778, 4 samples)",
                 kind="bulk", sample_sheet="samples.csv").save()
new.version                                     # 2: the next version

Track your code

One line at the top makes the script one run on baSeq: its code, the data it reads, what it adds and what it found. Reads are counted, not listed one by one; everything a run adds is kept.

bq.track()                                      # this script, as one run
counts = ds.files["counts.tsv"].load()          # noted as what the run read
bq.Record("dge_dex.csv").save()                 # a file of the run
bq.finish(finding="Dexamethasone raised ZBTB16, FKBP5, KLF15 and PER1 the most.",
          genes=["ZBTB16+", "FKBP5+", "KLF15+", "PER1+"])

In R:

bq_track()
Record$add("dge_dex.csv")
bq_finish(finding = "Dexamethasone raised ZBTB16 and FKBP5 the most.",
          genes = c("ZBTB16+", "FKBP5+"))

With an AI agent

An agent such as Claude Code can use baseqdb with your key. Ask it to start each task with baseqdb track and end it with baseqdb finish: everything it runs in between is one run.

baseqdb track "Dexamethasone response in Airway"
python filter_counts.py --min-reads 10          # joins the run
python dge_dex.py --padj 0.01                   # joins the run
baseqdb finish --finding "ZBTB16 and FKBP5 rose the most." \
               --gene ZBTB16+ --gene FKBP5+

Data you fetch with a key leaves baSeq. Our Data Processing Agreement does not cover what an AI agent or another service then does with it.