biomate-ai/biomate-bioconductor-kb/harmonizr
Claude Code and agent skill by biomate-ai/biomate-bioconductor-kb
- community
Checking that the SKILL.md file is still in the repository.
An implementation, which takes input data and makes it available for proper batch effect removal by ComBat or Limma. The implementation appropriately handles missing values by dissecting the input matrix into smaller matrices with sufficient data to feed the ComBat or limma algorithm. The adjusted data is returned to the user as a rebuild matrix. The implementation is meant to make as much data available as possible with minimal data loss.
Install
npx vskill@latest install biomate-ai/biomate-bioconductor-kb/harmonizrSKILL.md
The same SKILL.md was published earlier as biomate-ai/biomate-bioconductor-kb/harmonizr.
HarmonizR
Workflows
Standard Workflow
Perform batch effect correction on a data frame containing missing values using ComBat or limma.
Input: A data frame of features with potential missing values and a sample description data frame; Output: A batch-corrected data frame.
When to Use
- To perform batch effect correction on high-throughput biological datasets (especially proteomics) containing missing values using
harmonizR(). - To run ComBat or limma batch correction algorithms on dissected matrices to maximize feature rescue.
- To correct batch effects in a
SummarizedExperimentobject containing batch metadata incolData.
When NOT to Use
- For datasets with absolutely no missing values, you can use
sva::ComBatorlimma::removeBatchEffectdirectly becauseHarmonizR's matrix dissection overhead is unnecessary. - For non-log-transformed data where a log transformation is not appropriate, because
HarmonizRassumes log-transformed input for plotting and standard processing.
Data Requirements
- Input Format: A
data.frame,matrix, orSummarizedExperimentobject. - Structure: Features (e.g., proteins, genes) as rows and samples as columns. A separate description data frame mapping sample IDs to batch numbers is required unless using a
SummarizedExperimentwith batch info incolData. - Normalization State: Log-transformed continuous values (e.g., log2 intensity) are assumed and recommended.
Key Parameters
- data_as_input (NULL): Path to raw data or a data frame/matrix of features.
- description_as_input (NULL): Path to description file or a data frame mapping samples to batches.
- algorithm` ("ComBat"): Batch correction algorithm to use, either
"ComBat"or"limma". - ComBat_mode (1): Integer (1 to 4) specifying ComBat prior settings (e.g., parametric vs. non-parametric, mean-only).
- plot (NULL): Type of diagnostic plot to generate, e.g.,
"samplemeans","featuremeans", or"CV". - output_file ("cured_data"): Name or path of the output
.tsvfile. - cores (all available): Number of CPU cores to use for parallel execution.
- ur (TRUE): Logical indicating whether to apply unique removal of combinations for increased feature rescue.
Best Practices
- Keep the unique removal parameter
ur = TRUEenabled to maximize the rescue of features across batches. - Log-transform your data before running
harmonizR()to ensure correct behavior of the underlying ComBat/limma algorithms and plotting functions. - Use the
blockparameter to group batches together during matrix dissection to reduce runtime on large datasets.
Common Pitfalls
- Plotting un-logged data: Fix by log-transforming the input matrix prior to running
harmonizR()with plotting enabled. - Extremely long runtimes on datasets with many batches: Fix by setting the
blockparameter to an integer greater than 1 to pack batches together during dissection. - Missing batch metadata: Fix by ensuring the description data frame contains
ID,sample, andbatchcolumns matching the input matrix column names.
Alternatives
sva(specificallyComBat) for batch correction when there are no missing values.limma(specificallyremoveBatchEffect) for linear model-based batch correction without missing values.
Citations
- Schlumbohm et al. 2022, Nature Communications (for the HarmonizR framework).
References
- Homepage: bioconductor.org/packages/HarmonizR
- Vignette: https://bioconductor.org/packages/release/bioc/vignettes/HarmonizR/inst/doc/HarmonizR_Vignette.html
Security
- Security scanPassed, 100 of 100Pattern checks on the skill's files.
- Model reviewNot runA model review is scheduled for the next rescan.
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingskills/transcriptomics/harmonizr/SKILL.md
Popularity
- Repository stars
- 805
- Repository forks
- 67
- 7-day trend score
- 33
Stars and forks are for the whole GitHub repository, not this skill alone. Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/biomate-ai/biomate-bioconductor-kb/harmonizr)