Skip to content
BioMedsAI
Padma AI ScientistThe core of BioMedsAI · live

Padma, the AI scientist

Padma designs your study, runs the analysis when you press Start, generate graphs and interprets every result. Scientist-led analysis is there side by side, whenever you want a human scientist.

Start with PadmaOpen the platform ↗

Design review

sample_sheet.csv
Unit
Biological replicate
Comparison
Condition vs control
Replication
Replicates in every group
Confounding
A covariate aligned with the comparison

What this design cannot answer

Whether a difference comes from the condition or from the confounded covariate.

Padma proposes a fix before a single sample is sequenced.

Illustration.
Status

Open. A verified institutional or company email with a complete profile opens a free trial at once; personal email addresses (Gmail and similar) and full access open with a package.

What Padma does

Reviews the design

Experimental unit, replication, batch and aliased covariates, read from your sample sheet before you sequence.

Plans the analysis

Proposes the plan and settings; anything off-default is confirmed by you.

Runs on sealed engines

Starts on your Start, with the plan frozen and hashed and the run pinned to its image and seeds.

Interprets with a critic

Every claim tied to its table and row, checked against the evidence and the design.

Answers from your results

Queries whole tables and reads your figures; never states a number no tool produced.

Re-runs on request

Another resolution, a lineage focus, pseudobulk or more gene sets: a new plan with a new hash.

Briefs from the literature

Writes research briefs from a curated literature collection, with citations.

Hands over to a scientist

A BioMedsAI scientist is one message away, and can take the study on as a scientist-led project.

NVIDIA accelerated computing

Padma runs on NVIDIA accelerated computing.

Supported by NVIDIA through the NVIDIA Inception program, Padma brings accelerated computing to every study. Each analysis runs on a sealed engine in the cloud, and Padma interprets every result with you.

NVIDIA Inception Program member badge
  • DNA
  • RNA
  • Protein
  • Clinical data
NVIDIA

Accelerated computing

Padmasealed bundle

  • Report
  • Tables
  • Figures
  • Run record

What happens in a project.

  1. 1

    Project

    Name the study, state the question, choose the study.

  2. 2

    Upload

    Sample sheet, count matrices. Files go straight to an encrypted, per-project location.

  3. 3

    Design review

    Padma reads the sample sheet: experimental unit, arms, replication, batch, aliased covariates, and your inputs.

  4. 4

    Run

    Padma prepares the analysis. You press Start; the plan is frozen, hashed and run on a sealed engine.

  5. 5

    Delivery

    Report, tables, figures, claims and the run record arrive in the project as a sealed bundle.

  6. 6

    Questions

    Ask about any delivered table or figure, or ask for a re-run with different settings.

Watch a run. Open what it delivers.

padma › run my-study

Started by you
  1. intakeSample sheet read; libraries checksummed
  2. designPlan frozen against the design card
  3. qualityLow-quality cells and doublets removed
  4. integrationSamples integrated; cell types annotated
  5. expressionDifferential expression per cell type, condition vs control
  6. pathwaysGene-set and pathway enrichment
  7. claimsEach claim tied to its table and row
  8. criticClaims checked against the evidence and the design
  9. sealedBundle written with its run record and checksums
Illustrative trace. Step names follow the single-cell pipeline.

my-study/

report.html

Opens with the claims, the critic's verdict and what the analysis cannot show; then methods, figures and every table, each linked to the step that made it.

Illustrative bundle. Select a file to see what it holds.