Ranked annotation prioritization

ranked-annotation-prioritization

Summary

This skill reconstructs the diffusion prioritization stage of metabolomics annotation, which combines network-propagated diffusion scores with recovered peaks to build a final ranked compound annotation table. It is essential for converting cluster-filtered KEGG candidates into a context-aware, metabolite-network-informed ranking that prioritizes biologically plausible annotations.

When to use

Apply this skill when you have completed cluster-based filtering of KEGG candidate assignments in untargeted LC-MS metabolomics and need to rank those candidates by biological plausibility using a metabolite interaction network. Specifically, use it after clusterBased.filter has assigned quasi-molecular adducts to feature clusters, but before delivering a final annotated peak table to the user.

When NOT to use

• Input is already a fully annotated feature table or peak-intensity matrix; this skill applies only after cluster-based candidate filtering
• No KEGG metabolite network graph is available; diffusion propagation requires a connected network representation
• All peaks have been retained by cluster-based filtering and no recovery step is needed (skill is still applicable but the recovery step would be trivial)

Inputs

• Cluster-filtered candidate assignments (output from clusterBased.filter)
• Sample FELLA graph (igraph object representing KEGG metabolite network)
• Feature-to-cluster mapping with intensity information

Outputs

• Ranked.Tab: final ranked annotation table sorted by diffusion-prioritized scores
• Recovered peaks: peaks restored from complete removal by cluster-based filtering

How to apply

The diffusion prioritization workflow operates in five steps. First, compute diffusion input scores from the cluster-filtered candidates using the diffusion.input function with input.type set to either 'probability' (soft scores) or 'binary' (hard assignments). Second, execute set.diffusion on the FELLA sample.graph to propagate scores through the metabolite network, allowing metabolites connected to high-scoring candidates to inherit signal. Third, apply z-score normalization to the diffusion scores to account for network topology and degree bias (setting scores = z rather than raw). Fourth, use recoveringPeaks to restore peaks that were completely removed by cluster-based filtering, merging them back by compound identifier. Fifth, compile and rank the final results using the finalResults function with z-score normalization, producing the Ranked.Tab output sorted by diffusion score.

Related tools

• mWISE (R package providing diffusion.input, set.diffusion, recoveringPeaks, and finalResults functions for ranked prioritization) — https://dev.b2s.club/b2slab/mWISE
• FELLA (Provides the sample.graph metabolite network (igraph object) used for diffusion score propagation)
• igraph (Graph manipulation library; used to construct and traverse the undirected metabolite network)
• R (Execution environment for mWISE functions and statistical normalization (z-score calculation))

Examples

diffusion.scores <- set.diffusion(diffusion.input(candidates, input.type='probability'), sample.graph); recovered <- recoveringPeaks(filtered.candidates); final <- finalResults(merge(diffusion.scores, recovered, by='compound_id'), scores='z')

Evaluation signals

• Diffusion scores are computed for all cluster-filtered candidates and fall within a biologically plausible range (typically normalized to z-scores with mean 0, SD 1)
• Recovered peaks are successfully merged back into the final table by compound identifier with no duplicates or missing values
• The Ranked.Tab output is sorted in descending order by diffusion score, with the highest-scoring candidate per peak ranked first
• Z-score normalization has been applied; raw diffusion scores should NOT appear in the final output
• The final table schema includes columns for mass-to-charge ratio, cluster ID, compound identifier, diffusion score, and rank position

Limitations

• Diffusion prioritization depends entirely on the quality and completeness of the underlying KEGG metabolite network; missing or misannotated edges may bias propagation
• Peaks completely removed by cluster-based filtering are recovered, but their diffusion scores may be lower due to lack of supporting evidence in the filtered candidate set
• Z-score normalization assumes the distribution of diffusion scores approximates a normal distribution; severely skewed networks may produce unintuitive rankings
• The method is sensitive to the input.type parameter (probability vs. binary); choice of input mode can substantially affect final rankings

Evidence

• [intro] The diffusion prioritization stage operates by: (1) computing diffusion input from filtered candidates using diffusion.input; (2) applying set.diffusion with z-score normalization on the FELLA graph to obtain diffusion scores; (3) recovering completely removed peaks with recoveringPeaks; (4) merging the recovered peaks with diffusion results by compound identifier; and (5) building a final ranked table using finalResults function with z-score normalization.: "The diffusion prioritization stage operates by: (1) computing diffusion input from filtered candidates using diffusion.input; (2) applying set.diffusion with z-score normalization on the FELLA"
• [intro] The different diffusion inputs can be computed using the diffusion.input function. The input.type argument can be set to probability or binary.: "The different diffusion inputs can be computed using the diffusion.input function. The input.type argument can be set to probability or binary."
• [intro] The z score normalizes the diffusion scores by taking into account the topology of the graph. On the other hand, when scores = raw, no normalization is applied.: "The z score normalizes the diffusion scores by taking into account the topology of the graph. On the other hand, when scores = raw, no normalization is applied."
• [intro] The recoveringPeaks function recovers the peaks that have been completely removed by the cluster-based filter.: "The recoveringPeaks function recovers the peaks that have been completely removed by the cluster-based filter."
• [intro] mWISE integrates several strategies to provide a fast annotation of peak-intensity tables. It consists of three main steps aimed at i) matching mass-to-charge ratio values to KEGG database, ii) clustering and filtering the potential KEGG candidates, and building a final prioritized list using diffusion in networks: "mWISE integrates several strategies to provide a fast annotation of peak-intensity tables. It consists of three main steps aimed at i) matching mass-to-charge ratio values to KEGG database, ii)"