Skip to content

UC-8.1 — Minimal Sample Grouping for Complete Compound Coverage

Module: 8 – Assembly of Functional Consortia
Visualization type: Interactive faceted scatter plot (minimal non-redundant functional guilds)
Primary inputs: BioRemPP_Results.xlsx or BioRemPP_Results.csv (sample–compound associations)
Primary outputs: Minimal set of "functional guilds" providing complete compound coverage within a selected chemical class

Scientific Question and Rationale

Question: What is the minimum number of distinct compound co-annotation profiles (i.e., groups of samples with identical compound co-annotation sets) required to collectively cover all observed compound co-annotations within a given chemical class?

This use case formalizes candidate consortium selection as a set cover problem at the annotation level. By clustering samples into annotation groups that share identical compound co-annotation sets within a specific chemical class, and then selecting the smallest subset of these groups that collectively cover all compounds, the analysis may reveal the minimal, non-redundant candidate set needed to achieve full compound annotation coverage. This can provide a principled, annotation-based foundation for hypothesis-driven consortium design (experimental validation is required to confirm functional capacity).

Data and Inputs

  • Primary data source: BioRemPP_Results.xlsx or BioRemPP_Results.csv (semicolon-delimited)
  • Key columns:
  • sample – identifier for each biological sample
  • compoundclass – chemical class to which a compound belongs

  • compoundname – name of the chemical compound interacted with by the sample

  • Core concept: Annotation Groups An annotation group is defined as a group of samples that share the exact same set of compound co-annotations within a selected chemical class. Samples within the same group have identical annotation profiles with respect to that class.

Analytical Workflow

  1. User Selection (Chemical Class)
    The user selects a compoundclass from an interactive dropdown menu (e.g., Aromatics, Chlorinated, Metals). This choice defines the chemical space to be analyzed.

  2. Dynamic Filtering
    The primary results table (BioRemPP_Results.xlsx or BioRemPP_Results.csv) is filtered to retain only rows where:

  3. compoundclass equals the selected class, and
  4. both sample and compoundname are valid and non-missing.

This subset represents all observed compound co-annotations for samples within the chosen class.

  1. Sample Grouping into Annotation Groups For each sample, the set of associated compoundname values (within the selected class) is computed.
  2. Samples whose compound co-annotation sets are identical are grouped into the same annotation group.

  3. Each group is thus characterized by a unique compound annotation profile and a list of member samples.

  4. Minimization via Set Cover (Greedy Heuristic) A greedy set cover algorithm is applied to the annotation groups:

  5. Initialize the set of uncovered compounds as all unique compoundname values in the selected class.
  6. Iteratively select the group that covers the largest number of currently uncovered compounds.
  7. Remove these compounds from the uncovered set and repeat until all compounds are covered.

The result is a minimal (or near-minimal) subset of annotation groups that collectively cover every compound co-annotation in the selected class.

  1. Rendering
    The minimal guild set is visualized as a faceted scatter plot:
  2. each facet (subplot) corresponds to one selected functional guild,
  3. within a facet, the x-axis lists member samples and the y-axis lists compoundnames,
  4. points mark the presence of a sample–compound interaction, visually confirming the guild's profile.

How to Read the Plot

  • Dropdown Menu
    Use the dropdown to choose the Compound Class. The full pipeline (filtering, guild detection, set cover, and plotting) is recomputed and the visualization is updated accordingly.

  • Subplots (Facets) Each subplot represents one minimal required annotation group:

  • the number of subplots equals the number of annotation groups selected by the set cover algorithm,

  • each subplot represents a distinct compound co-annotation profile covering a subset of the selected class.

  • Y-axis (Compounds)
    Lists the Compound Names that define the chemical profile of that guild within the selected class.

  • X-axis (Samples)
    Lists the Samples that belong to that guild.
    All samples within a subplot share the same compound interaction profile.

  • Points (Co-annotations) A point at the intersection of a sample and a compound indicates an observed co-annotation (i.e., the sample has at least one KO annotation linking it to that compound).

Representative Output

The image below illustrates a representative output generated by this use case using the example dataset.

Click on the image to enlarge and explore details.

Representative output for UC-8.1

Interpretation and Key Messages

  • Annotation Redundancy The number of subplots (minimal annotation groups) can provide an immediate measure of annotation redundancy:
  • Few groups → high redundancy: many samples share similar compound co-annotation profiles.

  • Many groups → low redundancy: compound annotation coverage relies on multiple distinct annotation profiles.

  • Core vs. Niche Annotation Profiles

  • A group whose subplot contains many samples may represent a widely distributed co-annotation profile common across samples; these could be robust, broadly shared annotation patterns.

  • A group with only one or two samples may represent a niche annotation profile, potentially encoding co-annotations that cannot be easily replaced.

  • Annotation-guided Minimal Candidate Set This analysis can be valuable for identifying minimal candidate consortia at the annotation level:

  • by selecting one representative sample per annotation group from the minimal set, one could achieve complete compound annotation coverage for the chosen class,

  • this may minimize annotation overlap while preserving full coverage, useful for hypothesis generation (experimental validation required to confirm functional capacity).

  • Trade-offs and Extensions Although the minimal set emphasizes parsimony, additional samples from the same annotation groups can be added to increase annotation redundancy (e.g., as a proxy for potential robustness against variability or loss of specific members).

Reproducibility and Assumptions

  • Input Format
    The analysis assumes a semicolon-delimited BioRemPP_Results.xlsx or BioRemPP_Results.csv table containing at least:
  • sample,
  • compoundclass,

  • compoundname.

  • Definition of Annotation Groups Annotation groups are defined strictly by identical compound co-annotation sets within the selected class. Even a single additional or missing compound differentiates two groups.

  • Set Cover Approximation
    The minimization step uses a greedy heuristic for the set cover problem:

  • this approach is computationally efficient and typically near-optimal,

  • however, it does not guarantee the absolute mathematically minimal number of guilds in all possible cases.

  • Annotation Interpretation The analysis is based on observed sample–compound co-annotations. It does not encode kinetic parameters, environmental abundance, or interaction strength; it answers a structural annotation question: Which distinct co-annotation profiles are minimally required to cover all compound co-annotations in a given class?

Activity diagram of the use case

Click on the image to enlarge and explore details.

Activity diagram of the use case