Deterministic analysis · traceable evidence

Turn experimental data into
conclusions that withstand scrutiny.

Start with an expression matrix. Inspect the data, run differential analysis, verify the literature, test a candidate gap, and plan the next experiment. Tools calculate the statistics; every evidence claim opens to a source.

Start an analysis See how it works

CSV / TSV · resumable results · PMID / DOI evidence · reproducible code

Research question

Which genes changed reliably between treatment and control?

01
Inspect data

6 samples · control 3 / treatment 3 · no missing values

02
Differential expression

Welch t-test · BH FDR · 6 significant genes

03
Retrieve evidence

8 candidate papers · 3 highly relevant

Differential expression

Results you can inspect

6 significant
Genelog₂FCpadj
MYC+2.310.0012
IL6+1.840.0048
BRCA1−1.520.012
Verified · PMID 38411209Openable source evidence, not model memory

Mechanistic interpretations stay attached to retrieved records. Zero hits are marked as insufficient evidence.

Why CiteLab

Research improves through another check, not another paragraph.

One continuous workspace keeps data, methods, evidence, candidate gaps, and next actions in the same task.

01

Calculate before interpreting

Data checks, Welch tests, and BH FDR run in deterministic tools rather than improvised prose.

02

Open every source

Evidence connects to PubMed, OpenAlex, PMID, and DOI; zero hits stay explicitly insufficient.

03

Test the gap, then act

Separate support, counterevidence, uncertainty, feasibility, and the next experiment.

Built for real research work

Designed around work that actually happens at the bench.

Transcriptomic DEGGO / KEGG enrichmentLiterature evidenceResearch-gap validationOrthogonal validation plan

One continuous workflow

From data to the next experiment without losing context.

The task, results, evidence, search protocol, and follow-up questions remain in one research record.

  1. 01

    Inspect data

    Identify groups, sample size, missingness, and format risks.

  2. 02

    Run analysis

    Use explicit methods and thresholds to produce inspectable results.

  3. 03

    Build evidence

    Retrieve papers, preserve the query strategy, and show evidence boundaries.

  4. 04

    Validate the gap

    Compare support and counterevidence before planning the next experiment.

Scientific restraint by default

No result is not novelty. Association is not causation.

CiteLab shows methods, sample sizes, thresholds, warnings, and retrieval provenance. A model cannot fill in a result a tool did not obtain.

Read the analysis and evidence guides

Start with a real question

Keep the next analysis on one research bench.

Start free and run the complete workflow with demo data.

Create a workbench
CiteLab — Biomedical research workbench