feat(getSubnetworkFromIndra): Add correlation metrics into INDRA results

[tonywu1999]

PR Type

Enhancement, Documentation, Bug fix

---

Description

• Added correlation metrics to INDRA subnetwork results.

• Introduced new parameters for correlation filtering and protein-level data.

• Updated documentation to reflect new functionality and parameters.

• Fixed issues related to correlation handling in edge filtering.

---

Changes walkthrough 📝

	Relevant files
Dependencies	DESCRIPTION Added `tidyr` dependency to Imports.                                          DESCRIPTION Added tidyr to the Imports section. +2/-1      NAMESPACE Added imports for `MSstats` and `tidyr`.                                  NAMESPACE Imported quantification from MSstats. Imported pivot_wider from tidyr. +2/-0
Enhancement	getSubnetworkFromIndra.R Added correlation handling in `getSubnetworkFromIndra`.    R/getSubnetworkFromIndra.R Added protein_level_data and correlation_cutoff parameters. Updated function logic to handle correlation filtering. Enhanced edge filtering with correlation metrics. +12/-3    utils_getSubnetworkFromIndra.R Enhanced edge construction and filtering with correlation. R/utils_getSubnetworkFromIndra.R Modified .constructEdgesDataFrame to calculate and include correlations. Updated .filterEdgesDataFrame to filter edges by correlation cutoff. Added logic to handle protein-level data for correlation computation. +23/-3
Documentation	getSubnetworkFromIndra.Rd Updated documentation for new parameters and features.      man/getSubnetworkFromIndra.Rd Updated documentation to include protein_level_data and correlation_cutoff. Reflected new functionality in parameter descriptions. +12/-1

---

💡 PR-Agent usage: Comment /help "your question" on any pull request to receive relevant information

[github-actions]

PR Reviewer Guide 🔍

Here are some key observations to aid the review process:

⏱️ Estimated effort to review: 4 🔵🔵🔵🔵⚪

🧪 No relevant tests

🔒 No security concerns identified

⚡ Recommended focus areas for review Correlation Calculation Logic The correlation calculation in .constructEdgesDataFrame assumes that the wide_data matrix is correctly structured and that the cor function will not encounter issues with missing or invalid data. This logic should be validated for edge cases, such as missing values or mismatched protein IDs. if (!is.null(protein_level_data)) { protein_level_data <- protein_level_data[ protein_level_data$Protein %in% edges$source | protein_level_data$Protein %in% edges$target, ] wide_data = pivot_wider(protein_level_data[,c("Protein", "LogIntensities", "originalRUN")], names_from = Protein, values_from = LogIntensities) wide_data <- wide_data[, -which(names(wide_data) == "originalRUN")] correlations = cor(wide_data, use = "pairwise.complete.obs") edges$correlation = apply(edges, 1, function(edge) correlations[edge["source"], edge["target"]]) Filtering Logic for Correlation The filtering logic in .filterEdgesDataFrame for correlation cutoffs assumes that the correlation column is always present when protein_level_data is provided. This assumption should be validated to avoid runtime errors. if ("correlation" %in% colnames(edges)) { edges <- edges[which(abs(edges$correlation) >= correlation_cutoff), ] }

[github-actions]

PR Code Suggestions ✨

Explore these optional code suggestions:

Category	Suggestion	Score
Possible issue	Handle missing protein indices in correlation matrix Ensure that the correlations matrix is properly indexed and handles cases where source or target proteins are missing or mismatched in the matrix to avoid runtime errors. R/utils_getSubnetworkFromIndra.R [192] -edges$correlation = apply(edges, 1, function(edge) correlations[edge["source"], edge["target"]]) +edges$correlation = apply(edges, 1, function(edge) { + if (edge["source"] %in% rownames(correlations) && edge["target"] %in% colnames(correlations)) { + return(correlations[edge["source"], edge["target"]]) + } else { + return(NA) + } +}) Suggestion importance[1-10]: 9 Why: This suggestion addresses a potential runtime error by ensuring that the correlation matrix is properly indexed and handles cases where source or target proteins are missing. This is a critical improvement for robustness and correctness.	9
	Validate input structure for pivot_wider Ensure that the pivot_wider function call handles cases where the input data does not have the expected structure, such as missing required columns, to prevent runtime errors. R/utils_getSubnetworkFromIndra.R [189] +if (!all(c("Protein", "LogIntensities", "originalRUN") %in% colnames(protein_level_data))) { + stop("protein_level_data must contain 'Protein', 'LogIntensities', and 'originalRUN' columns.") +} wide_data = pivot_wider(protein_level_data[,c("Protein", "LogIntensities", "originalRUN")], names_from = Protein, values_from = LogIntensities) Suggestion importance[1-10]: 9 Why: This suggestion ensures that the input data for pivot_wider has the expected structure, preventing runtime errors and improving the function's robustness.	9
	Validate wide_data for missing values Add a check to ensure that the wide_data variable does not contain columns with all missing values before calculating correlations, as this could lead to errors or invalid results. R/utils_getSubnetworkFromIndra.R [191] +if (any(colSums(!is.na(wide_data)) == 0)) { + stop("wide_data contains columns with all missing values, unable to calculate correlations.") +} correlations = cor(wide_data, use = "pairwise.complete.obs") Suggestion importance[1-10]: 8 Why: Adding a check for columns with all missing values in wide_data prevents errors during correlation computation, enhancing the reliability of the function.	8
General	Validate edges after correlation filtering Add a safeguard to ensure that the edges data frame does not contain invalid or missing values after filtering by correlation, as this could lead to downstream issues. R/utils_getSubnetworkFromIndra.R [228] edges <- edges[which(abs(edges$correlation) >= correlation_cutoff), ] +if (nrow(edges) == 0) { + warning("No edges remain after applying the correlation cutoff.") +} Suggestion importance[1-10]: 7 Why: Adding a safeguard to check for empty edges after filtering by correlation cutoff is a good practice to avoid downstream issues, though the impact is less critical compared to other suggestions.	7

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 29.62963% with 19 lines in your changes missing coverage. Please review.

Project coverage is 83.58%. Comparing base (fbac585) to head (69ce251).

Files with missing lines	Patch %	Lines
R/utils_getSubnetworkFromIndra.R	20.83%	19 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##            devel      #32      +/-   ##
==========================================
- Coverage   87.09%   83.58%   -3.51%     
==========================================
  Files           7        7              
  Lines         434      457      +23     
==========================================
+ Hits          378      382       +4     
- Misses         56       75      +19     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.