Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

Bio Experimental Design Sample Size · FreedomIntelligence | DecimalAI

Bio Experimental Design Sample Size

SkillScore82Maintained

bio-experimental-design-sample-size

Estimates required sample sizes for differential expression, ChIP-seq, methylation, and proteomics studies. Use when budgeting experiments, writing grant proposals, or determining minimum replicates needed to achieve statistical significance for expected effect sizes.

scienceby FreedomIntelligence

Links this skill to your agents and keeps it current with the author’s updates.

Automatic

Research pack

1 · One-time — install the CLI (docs ↗)

pip install decimalai

2 · Get the skill — writes .claude/skills/bio-experimental-design-sample-size/SKILL.md

decimalai skills pull bio-experimental-design-sample-size --out .claude/skills/

No CLI? Get the same file — curl · raw file

curl -fsSL --create-dirs -o .claude/skills/bio-experimental-design-sample-size/SKILL.md https://app.decimal.ai/s/bio-experimental-design-sample-size/SKILL.md

Raw file: /s/bio-experimental-design-sample-size/SKILL.md ↗ · machine-readable summary at /s/bio-experimental-design-sample-size.json ↗

In productionGlobal

how this skill behaves on real traffic · last 30 days

Runs

all versions · 30d

Eval pass

100%

global

Used when offered

82%

used / offered · all versions

By model· real traffic, every org · 30d — not the benchmark

Model	Eval pass	Runs
gemini-3.6-flashlowest	95%	21
gemini-3.1-pro-preview	100%	4

Anonymized aggregate across every org using this skill — proof from real production use, not a demo.

Does this skill help? · Benchmark: with vs. withoutsign in to run this on your agents

22 test cases from the publisher’s eval.yaml · each run twice on gemini-3.6-flash — with the skill injected, then without · last run 50d ago

Yes — +36% lift-14% tokens0% turns91% of whole cases pass with skill (20/22)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-21	✗→✓	▲ Improved	-38%	0%
case-01	✗→✓	▲ Improved	-4%	0%
case-03	✗→✓	▲ Improved	6%	0%
case-10	✗→✓	▲ Improved	-48%	0%
case-11	✗→✓	▲ Improved	-29%	0%

Showing 5 of 22 cases

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/1/2026

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

FreedomIntelligence / OpenClaw-Medical-Skills

skills/bio-experimental-design-sample-size/SKILL.md

View on GitHub ↗

Imported 8/1/2026Version v2

Skill contents

SKILL.md

<!--

COPYRIGHT NOTICE

This file is part of the "Universal Biomedical Skills" project.

Copyright (c) 2026 MD BABU MIA, PhD <md.babu.mia@mssm.edu>

All Rights Reserved.

This code is proprietary and confidential.

Unauthorized copying of this file, via any medium is strictly prohibited.

Provenance: Authenticated by MD BABU MIA

-->

Sample Size Estimation

RNA-seq Sample Size

r
library(ssizeRNA)

# Estimate sample size for RNA-seq
# m = total genes, m1 = expected DE genes
# fc = fold change, fdr = target FDR
result <- ssizeRNA_single(nGenes = 20000, pi0 = 0.9, m = 200,
                          mu = 10, disp = 0.1, fc = 2,
                          fdr = 0.05, power = 0.8)
result$ssize  # Required n per group

DESeq2-based Estimation

r
library(DESeq2)

# From pilot data
dds_pilot <- DESeqDataSetFromMatrix(pilot_counts, colData, ~condition)
dds_pilot <- DESeq(dds_pilot)

# Extract dispersion estimates for power calculation
dispersions <- mcols(dds_pilot)$dispGeneEst
median_disp <- median(dispersions, na.rm = TRUE)
# Use median_disp in power calculations

Single-cell Sample Size

r
library(powsimR)

# Estimate for scRNA-seq
# Accounts for dropout and cell-to-cell variability
params <- estimateParam(pilot_sce)
power <- simulateDE(params, n1 = 100, n2 = 100,
                    p.DE = 0.1, pLFC = 1)

Sample Size by Assay Type

| Assay | Min Recommended | For Small Effects | |-------|-----------------|-------------------| | Bulk RNA-seq | 3 | 6-12 | | scRNA-seq | 3 samples, 1000 cells | 6+ samples | | ATAC-seq | 2 | 4-6 | | ChIP-seq | 2 | 3-4 | | Proteomics | 3 | 6-10 | | Methylation | 4 | 8-12 |

Budget Optimization

When resources are limited, prioritize:

Biological replicates over technical replicates
More samples over deeper sequencing (after ~20M reads for RNA-seq)
Balanced designs (equal n per group)

Related Skills

experimental-design/power-analysis - Power calculations
experimental-design/batch-design - Optimal batch assignment
single-cell/preprocessing - scRNA-seq experimental design

Illustrative Want to see what bio-experimental-design-sample-size would put in front of your model? Preview the rendered body and check its trigger phrases — no LLM call, no signup.

Latest benchmark · publisher runVerified

8/2/2026, 16:27 UTC · 22 cases

Lift

+36%

91% vs 55%

Agent turns

1.0 vs 1.0

Tokens

-14%

2,155.77 vs 2,496.77

fail → pass · 8pass → pass · 12fail → fail · 2

Per-case results (22)

Case	Status	Duration (ms)			Turns			Tokens			Tool calls
Case	Status	Without	With	Δ	Without	With	Δ	Without	With	Δ	Without	With	Δ
case-05 Our bulk RNA-seq study was sequenced across two distinct library preparation batches. How do I apply ComBat-seq from the sva package in R to correct integer read counts before downstream differential expression?	pass→pass	13,179	10,595	-20%	1	1	0%	2,422	2,767	+14%	0	0	—
case-06 I just received raw FASTQ files from Illumina paired-end sequencing. What is the recommended command-line syntax for running Fastp or Trimmomatic to trim low-quality bases below Phred 20 and remove adapter contamination?	pass→pass	11,490	9,930	-14%	1	1	0%	2,304	2,834	+23%	0	0	—
case-21 Compare the recommended minimal sample size versus small effect sample size specifically for ATAC-seq versus mass spectrometry Proteomics assays.	fail→pass	17,348	8,181	-53%	1	1	0%	3,148	1,956	-38%	0	0	—
case-01 I am planning a bulk RNA-seq experiment and have a small pilot dataset processed in DESeq2. Could you show me how to extract the gene-wise dispersion estimates from my DESeq2 object and use `ssizeRNA` in R to calculate the required sample size per group for target statistical power and FDR? Please provide the R code workflow.	fail→pass	37,020	13,696	-63%	1	1	0%	3,620	3,480	-4%	0	0	—
case-02 We are designing a single-cell RNA sequencing study and want to perform a power simulation using pilot data. Can you demonstrate how to use `powsimR` in R to estimate sample size and cell requirements while accounting for dropout, along with key advice on allocating budget between biological replicates and read depth?	fail→fail	18,538	18,990	+2%	1	1	0%	3,566	4,394	+23%	0	0	—
case-03 I need a benchmark summary of recommended sample sizes across various genomic assays—such as bulk RNA-seq, scRNA-seq, ATAC-seq, ChIP-seq, and DNA methylation—comparing standard minimal replicate counts with studies targeting small effect sizes. Please also outline general principles for resource optimization when sequencing budgets are constrained.	fail→pass	22,475	19,718	-12%	1	1	0%	3,996	4,237	+6%	0	0	—
case-04 I have a completed bulk RNA-seq matrix with 6 samples per condition and want to run differential expression testing using DESeq2 to identify significant genes at FDR < 0.05. How should I set up the design matrix, run the DESeq pipeline, and shrink log2 fold changes using apeglm?	pass→pass	12,235	8,352	-32%	1	1	0%	2,471	2,568	+4%	0	0	—
case-07 We want to calculate sample sizes for an R-based bulk RNA-seq power calculation. What specific parameters are required for the single-dataset ssizeRNA calculation when testing 20,000 genes with a target proportion of non-differential genes of 0.9, fold change of 2, FDR 0.05, and 80% power?	fail→fail	20,014	7,244	-64%	1	1	0%	2,720	2,199	-19%	0	0	—
case-08 We are designing a mass spectrometry proteomics experiment and considering using 2 replicates per condition to save costs. Is 2 sufficient, and how many replicates are recommended for standard detection versus detecting subtle/small effect size changes?	pass→pass	15,222	9,192	-40%	1	1	0%	2,595	2,185	-16%	0	0	—
case-09 Our lab wants to sequence a single-cell RNA-seq experiment with 1 biological replicate per group but 10,000 cells per sample. Is 1 sample with high cell count adequate according to standard experimental design guidelines, or what minimum sample and cell count per sample is recommended?	pass→pass	11,910	6,039	-49%	1	1	0%	2,090	1,756	-16%	0	0	—
case-10 What is the baseline minimum sample size requirement for ATAC-seq chromatin accessibility profiling, and how many replicates should be used if expecting subtle open-chromatin differences between closely related cell sub-states?	fail→pass	12,751	3,074	-76%	1	1	0%	2,347	1,225	-48%	0	0	—
case-11 We are planning a histone modification ChIP-seq experiment. Standard protocols often suggest 2 replicates, but we expect weak binding changes across conditions. What is the minimum recommended replicate count and the adjusted target for small effects?	fail→pass	13,291	4,444	-67%	1	1	0%	2,094	1,490	-29%	0	0	—
case-12 We are designing a bisulfite sequencing DNA methylation study. Is 2 replicates per group enough for genome-wide methylation profiling, or what is the recommended minimum and small-effect sample size?	pass→pass	15,281	6,307	-59%	1	1	0%	2,587	1,720	-34%	0	0	—
case-13 For standard bulk RNA-sequencing across two treatment groups, what is the minimum baseline biological replicate count per group, and how many replicates are recommended if detecting small fold changes?	pass→pass	10,209	3,209	-69%	1	1	0%	1,871	1,198	-36%	0	0	—
case-14 We have a fixed budget for an upcoming bulk RNA-seq study. Our technician suggests spending part of the budget running technical replicates for each sample. Should we invest in technical replicates or biological replicates?	pass→pass	11,468	4,336	-62%	1	1	0%	1,749	1,301	-26%	0	0	—
case-15 We are deciding whether to sequence our RNA-seq samples at 50 million reads per sample or increase our biological sample count from n=3 to n=6 while sequencing at 20 million reads per sample. What is the recommended read depth threshold after which adding more samples is prioritized over deeper sequencing?	pass→pass	11,612	6,171	-47%	1	1	0%	2,183	1,761	-19%	0	0	—
case-16 We have budget for 10 biological samples total in a two-arm comparative study (control vs treatment). A researcher suggests preparing 8 control samples and 2 treatment samples because control tissue is easier to acquire. What design principle should be followed regarding sample distribution across groups?	pass→pass	9,473	8,851	-7%	1	1	0%	1,719	2,238	+30%	0	0	—
case-17 When estimating dispersion parameters from pilot DESeq2 data for power calculations, should I extract gene dispersions using dispersion() or mcols(), and what summary statistic should be calculated across genes?	fail→pass	14,803	7,090	-52%	1	1	0%	2,622	1,830	-30%	0	0	—
case-18 In single-cell power simulation using pilot SingleCellExperiment data, what exact R syntax is used to estimate dataset parameters and subsequently simulate differential expression with group sizes n1=100 and n2=100, expected DE proportion 0.1, and log fold change 1?	fail→pass	12,850	4,555	-65%	1	1	0%	2,511	1,637	-35%	0	0	—
case-19 Show the concise R syntax using single-dataset ssizeRNA function to calculate sample size for 20,000 genes, mean read count 10, dispersion 0.1, fold change 2, target FDR 0.05, and 80% power. How is the required sample size per group extracted from the output object?	pass→pass	7,937	5,397	-32%	1	1	0%	1,793	1,910	+7%	0	0	—
case-20 We are planning a single-cell RNA sequencing experiment on primary human immune cells where subtle transcriptional shifts between patient cohorts are expected. What minimum sample size (number of biological samples) is recommended for small effects in scRNA-seq?	pass→pass	27,511	5,575	-80%	1	1	0%	2,432	1,701	-30%	0	0	—
case-22 What are the three core budget optimization rules for biomedical sequencing design when experimental resources are limited?	fail→pass	14,276	2,507	-82%	1	1	0%	2,089	1,040	-50%	0	0	—

These cases were also screened by the skill’s intent / safety pass.See the Trust review ↗Reproduce this benchmark in Colab ↗

How this skill was measured

DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +36 percentage points is the difference between those two pass rates over the 22 comparable cases.

Model measured: gemini-3.6-flash
Graded by: gemini-3.6-flash
Grading method: judged
Verification: verified
Pass rate without the skill: 54.6%
Pass rate with the skill: 90.9%
Cases attempted: 22
Cases counted toward lift: 22
Cases that regressed: 0
Skill version measured: v2
Run completed: 8/2/2026, 16:27 UTC

Measured example — same prompt, with vs. without the skill · from the benchmark run · case case-21

Prompt

Compare the recommended minimal sample size versus small effect sample size specifically for ATAC-seq versus mass spectrometry Proteomics assays.

Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.

Show both answers

SkillScore blends 4 signals3 of 4 complete

BenchmarkPass rate WITH skill on a verified run91

Live evalPass rate on real evaluated traces100%

AI ratingNeeds ≥10 rated traces

AdoptionBreadth of real use across orgs40

Efficiency · separate axis (at equal correctness)+0% turns-9% tokens

Shown for transparency; not credited unless a verified run shows no correctness regression.

Backed by 3 of 4 signals — each added signal makes the score more robust.

Verified on gemini-3.6-flash · re-tested 50 days ago

Model	Method	Date	Lift
gemini-3.6-flash	verified	7/24/2026	+36%

Other measured skills in the registry, with their headline benchmark lift.