MixMashNet: An R Package for Single and Multilayer Networks

Adrien Perigord; Alice Margherita Ornago; Caterina Gregorio; Davide Liborio Vetrano; Federico Triolo; Maria De Martino

arxiv: 2602.05716 · v3 · submitted 2026-02-05 · 📊 stat.ME · stat.CO

MixMashNet: An R Package for Single and Multilayer Networks

Maria De Martino , Federico Triolo , Adrien Perigord , Alice Margherita Ornago , Davide Liborio Vetrano , Caterina Gregorio This is my paper

Pith reviewed 2026-05-16 07:03 UTC · model grok-4.3

classification 📊 stat.ME stat.CO

keywords mixed graphical modelsmultilayer networksR packagenetwork estimationbootstrap resamplingcommunity stabilitymixed data types

0 comments

The pith

MixMashNet is an R package that estimates single and multilayer networks from mixed continuous, count, and categorical variables using Mixed Graphical Models.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper presents MixMashNet as a unified tool for building network models where variables of different types can interact within and across layers. Users specify the overall layer structure in advance, after which the package fits the model, measures edge uncertainty through resampling, and evaluates how stable the detected node groups are. This matters for analyses that combine survey data with measurements or other heterogeneous sources, because it avoids forcing all variables into the same category. The package also supplies community scores and interactive displays to make the resulting networks easier to interpret.

Core claim

MixMashNet provides an integrated framework for estimating and analyzing single and multilayer networks using Mixed Graphical Models (MGMs), accommodating continuous, count, and categorical variables. In the multilayer setting, layers may comprise different types and numbers of variables, and users can explicitly impose a predefined multilayer topology. Bootstrap procedures are implemented to quantify sampling uncertainty for edge weights and node-level centrality indices. The package includes tools to assess the stability of node community membership and to compute community scores that summarize the latent dimensions identified through network clustering, along with interactive Shiny apps.

What carries the argument

Mixed Graphical Models (MGMs) extended to user-specified multilayer topologies, with bootstrap resampling for edge and centrality uncertainty plus community stability checks.

If this is right

Analysts can fit networks directly to data containing both numeric measurements and categorical responses without recoding everything to one type.
Bootstrap intervals give concrete ranges for how strongly two nodes are connected and how central each node is.
Node groups can be checked for consistency across resamples so that reported communities are not artifacts of one sample.
Community scores condense the clustering output into numeric summaries of each latent dimension.
Shiny interfaces let users explore the fitted network interactively without writing additional code.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The package could be applied to combine psychological survey scales with physiological sensor readings in a single network model.
Researchers might compare multilayer structures across different populations by fixing the same topology and examining differences in estimated parameters.
Future extensions could add time-indexed layers to track how mixed-variable networks evolve.
The bootstrap and stability tools provide a practical way to report uncertainty that many existing network packages omit.

Load-bearing premise

Mixed Graphical Models correctly recover the true dependencies among mixed variable types and a user-chosen multilayer topology introduces no systematic bias.

What would settle it

A simulation in which data are generated from a known mixed-variable network with a fixed layer structure, then the package recovers substantially different edges, centrality values, or community assignments than the generating model.

read the original abstract

The R package MixMashNet provides an integrated framework for estimating and analyzing single and multilayer networks using Mixed Graphical Models (MGMs), accommodating continuous, count, and categorical variables. In the multilayer setting, layers may comprise different types and numbers of variables, and users can explicitly impose a predefined multilayer topology. Bootstrap procedures are implemented to quantify sampling uncertainty for edge weights and node-level centrality indices. In addition, the package includes tools to assess the stability of node community membership and to compute community scores that summarize the latent dimensions identified through network clustering. MixMashNet also offers interactive Shiny applications to support exploration, visualization, and interpretation of the estimated networks.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

MixMashNet bundles mixed graphical models into one R package for single and multilayer networks with mixed data types, bootstrap, and Shiny tools, but offers little evidence on whether user-imposed topologies introduce bias.

read the letter

The main thing to know is that this paper presents MixMashNet, an R package that puts mixed graphical models into a single toolkit for both single-layer and multilayer networks, handling continuous, count, and categorical variables together. Users can impose a predefined multilayer topology, run bootstrap for edge weights and centrality, compute community scores, and use Shiny apps for visualization and interpretation. That integration is the actual new piece: it saves practitioners from stitching together separate tools or writing custom code for mixed-variable network estimation. The multilayer option with explicit topology specification stands out as practical for applied settings where some layer structure is already known. The bootstrap and community features are standard but useful additions in one place. The soft spot is the absence of any simulation results or sensitivity checks on the imposed-topology feature. If the user-specified structure does not match the data-generating process, regularization for mixed variables could distort edge estimates and downstream scores, yet the description supplies no tests of that risk or guidance on when the assumption holds. Without those checks or concrete performance numbers, it is hard to judge how reliable the outputs are in practice. This paper is for applied statisticians and network analysts who work with heterogeneous data and want ready software rather than method development. A reader needing a convenient implementation to explore real datasets would get direct value from trying the package. It deserves peer review so referees can examine the code, request added validation simulations, and confirm the implementation matches the claims.

Referee Report

2 major / 2 minor

Summary. The manuscript describes MixMashNet, an R package implementing an integrated framework for single- and multilayer network estimation and analysis via Mixed Graphical Models (MGMs) that accommodate continuous, count, and categorical variables. Layers may differ in variable type and number; users may impose a predefined multilayer topology. The package supplies bootstrap procedures for edge-weight and centrality uncertainty, tools for node-community stability assessment and community-score computation, and interactive Shiny applications for visualization and interpretation.

Significance. If the implementation is correct and the topology-imposition feature is shown not to introduce systematic bias, the package would provide a practical, accessible tool for applied researchers working with mixed-type multilayer networks in domains such as genomics and social science. The inclusion of bootstrap uncertainty quantification and community-stability diagnostics adds immediate utility beyond existing MGM implementations.

major comments (2)

[§3.2] §3.2 (Multilayer topology imposition): The central claim that the package supplies a 'valid integrated framework' rests on the assumption that a user-specified multilayer topology does not systematically distort edge weights, bootstrap intervals, or community scores when it deviates from the data-generating structure. No simulation studies or sensitivity analyses are reported that test this assumption under mixed-variable regularization; this is load-bearing for the validity claim.
[Methods] Methods (MGM estimation for multilayer case): The manuscript does not specify how the mixed graphical model penalty or regularization is adapted when layers contain unequal numbers or types of variables and when a topology is imposed; without these details it is impossible to verify that the estimator remains consistent with standard MGM theory.

minor comments (2)

[Abstract] Abstract: the phrase 'community scores that summarize the latent dimensions identified through network clustering' is used without a brief definition or reference to the exact formula implemented in the package.
[Title/Abstract] The package name 'MixMashNet' appears inconsistently capitalized in the title versus the abstract text.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive and detailed review of the MixMashNet manuscript. We address each major comment below and describe the revisions we will make to strengthen the paper.

read point-by-point responses

Referee: [§3.2] §3.2 (Multilayer topology imposition): The central claim that the package supplies a 'valid integrated framework' rests on the assumption that a user-specified multilayer topology does not systematically distort edge weights, bootstrap intervals, or community scores when it deviates from the data-generating structure. No simulation studies or sensitivity analyses are reported that test this assumption under mixed-variable regularization; this is load-bearing for the validity claim.

Authors: We agree that the manuscript does not currently include simulation studies or sensitivity analyses examining the consequences of imposing a multilayer topology that deviates from the true data-generating structure. The topology-imposition option is presented as a user-controlled feature for incorporating domain knowledge, analogous to structural constraints in other graphical model implementations. To address the concern, we will add a new subsection to §3.2 containing targeted simulation experiments under mixed-variable regularization. These will systematically vary the degree of mismatch between imposed and true topologies and report resulting bias and coverage in edge weights, bootstrap intervals, and community scores, thereby providing empirical support for the validity claim. revision: yes
Referee: [Methods] Methods (MGM estimation for multilayer case): The manuscript does not specify how the mixed graphical model penalty or regularization is adapted when layers contain unequal numbers or types of variables and when a topology is imposed; without these details it is impossible to verify that the estimator remains consistent with standard MGM theory.

Authors: We acknowledge that the Methods section lacks explicit detail on how the MGM penalty is adapted for multilayer networks with unequal layer sizes, heterogeneous variable types, and imposed topologies. The implementation extends the standard MGM penalized likelihood by applying layer-specific regularization parameters (tuned via EBIC or cross-validation) and by masking the precision matrix according to the user-specified topology. We will revise the Methods section to include the full penalized log-likelihood formulation, the handling of mixed variable types through appropriate link functions, and the precise mechanism by which the imposed topology constrains the edge set. This will make the estimator's relationship to existing MGM theory transparent and verifiable. revision: yes

Circularity Check

0 steps flagged

No derivation chain; package description implements established MGM methods

full rationale

The manuscript is a software package description for MixMashNet. It presents no original mathematical derivations, equations, or first-principles predictions. All core functionality (MGM estimation for mixed variables, multilayer topology imposition, bootstrap uncertainty, community stability) is described as building on existing Mixed Graphical Model literature rather than deriving new results from the package's own fitted outputs or self-citations. No load-bearing step reduces by construction to a fitted parameter or self-referential definition. The central claim is simply that the package integrates and exposes these standard tools, which is not circular.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Based on the abstract alone, no free parameters, axioms, or invented entities are explicitly introduced; the work rests on standard mixed graphical model assumptions from prior literature.

pith-pipeline@v0.9.0 · 5421 in / 1058 out tokens · 32566 ms · 2026-05-16T07:03:52.146740+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Foundation/RealityFromDistinction.lean reality_from_one_distinction unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

The R package MixMashNet provides an integrated framework for estimating and analyzing single and multilayer networks using Mixed Graphical Models (MGMs), accommodating continuous, count, and categorical variables... users can explicitly impose a predefined multilayer topology.
IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

Estimation is performed through a nodewise approach... penalized likelihood, with the regularization parameter selected through cross-validation (CV) or the extended Bayesian information criterion (EBIC)

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.