BayesMallows: An R Package for the Bayesian Mallows Model

Sørensen, Øystein; Crispino, Marta; Liu, Qinghua; Vitelli, Valeria

doi:10.32614/rj-2020-026

Cited by 10 publications

(13 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Enumeration for n ≤ 9 When the number of items n is small, it is possible to enumerate all possible rankings {i 1 , ..., i n } ∈ P n , and compute the KL-divergence between the Pseudo-Mallows distribution and the Mallows posterior. We first generate full ranking datasets with n items and N users by drawing N independent rankings from the Mallows distribution with α 0 and ρ 0 using the BayesMallows R package (Sørensen et al, 2020). For convenience, we fix ρ 0 = {1, 2, ..., n}.…”

Section: Empirical Study Of the Optimal Pseudo-mallows Conjecturementioning

confidence: 99%

“…For each dataset, we then run the Mallows MCMC and the Pseudo-Mallows, as described in Algorithm 3, for a grid of different numbers of iterations. After running the algorithms, we obtain a point estimate for the consensus parameter ρ by calculating the CP consensus (Sørensen et al, 2020) based on the samples obtained by the algorithm. Then we record the footrule distance between the CP consensus and the truth ρ 0 in order to assess the estimation accuracy of both algorithms for a given computing time.…”

Section: Infer ρ From Full Ranking Datamentioning

confidence: 99%

“…One can achieve this by using a mixture of Mallows, i.e., assign users to clusters such that, within each cluster, the users' rankings follow the same Mallows distribution. In the case study described in Section 4.3.1, we used the Bayesian Mallows mixture (Vitelli et al, 2018;Liu et al, 2019b), available through the BayesMallows package (Sørensen et al, 2020), but computationally slow. Faster clustering algorithms, such as K-means clustering, do not always produce enough within-cluster homogeneity.…”

Section: Pseudo Mallowsmentioning

confidence: 99%

See 2 more Smart Citations

Pseudo-Mallows for Efficient Probabilistic Preference Learning

Liu¹,

Vitelli²,

Mannino³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

We propose the Pseudo-Mallows distribution over the set of all permutations of n items, to approximate the posterior distribution with a Mallows likelihood. The Mallows model has been proven to be useful for recommender systems where it can be used to learn personal preferences from highly incomplete data provided by the users. Inference based on MCMC is however slow, preventing its use in real time applications. The Pseudo-Mallows distribution is a product of univariate discrete Mallows-like distributions, constrained to remain in the space of permutations. The quality of the approximation depends on the order of the n items used to determine the factorization sequence. In a variational setting, we optimise the variational order parameter by minimising a marginalized KL-divergence. We propose an approximate algorithm for this discrete optimization, and conjecture a certain form of the optimal variational order that depends on the data. Empirical evidence and some

show abstract

Section: Empirical Study Of the Optimal Pseudo-mallows Conjecturementioning

confidence: 99%

Section: Infer ρ From Full Ranking Datamentioning

confidence: 99%

Section: Pseudo Mallowsmentioning

confidence: 99%

See 1 more Smart Citation

Pseudo-Mallows for Efficient Probabilistic Preference Learning

Liu¹,

Vitelli²,

Mannino³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…To perform inference, Vitelli et al 17 proposed a MCMC algorithm based on a Metropolis-Hastings (MH) scheme (see Sørensen et al 18 for details on the implementation).…”

Section: Bmm For Complete Datamentioning

confidence: 99%

“…Our current work is based on the Bayesian Mallows model (BMM) 17 with implementing solution BayesMallows. 18 BayesMallows already provides a computationally feasible inferential approach for the most important choices of distance among permutations, and it shows good accuracy when compared to competitors on datasets of moderate size. 19 However, for this method to be applicable to the typical data dimensions in -omics applications, variable selection is crucial.…”

Section: Introductionmentioning

confidence: 99%

Rank‐based Bayesian variable selection for genome‐wide transcriptomic analyses

Emilie

Fleischer

Vitelli

2022

Statistics in Medicine

Self Cite

View full text Add to dashboard Cite

Variable selection is crucial in high‐dimensional omics‐based analyses, since it is biologically reasonable to assume only a subset of non‐noisy features contributes to the data structures. However, the task is particularly hard in an unsupervised setting, and a priori ad hoc variable selection is still a very frequent approach, despite the evident drawbacks and lack of reproducibility. We propose a Bayesian variable selection approach for rank‐based unsupervised transcriptomic analysis. Making use of data rankings instead of the actual continuous measurements increases the robustness of conclusions when compared to classical statistical methods, and embedding variable selection into the inferential tasks allows complete reproducibility. Specifically, we develop a novel extension of the Bayesian Mallows model for variable selection that allows for a full probabilistic analysis, leading to coherent quantification of uncertainties. Simulation studies demonstrate the versatility and robustness of the proposed method in a variety of scenarios, as well as its superiority with respect to several competitors when varying the data dimension or data generating process. We use the novel approach to analyze genome‐wide RNAseq gene expression data from ovarian cancer patients: several genes that affect cancer development are correctly detected in a completely unsupervised fashion, showing the usefulness of the method in the context of signature discovery for cancer genomics. Moreover, the possibility to also perform uncertainty quantification plays a key role in the subsequent biological investigation.

show abstract

Transcriptomic pan‐cancer analysis using rank‐based Bayesian inference

et al. 2023

Self Cite

View full text Add to dashboard Cite

The analysis of whole genomes of pan‐cancer data sets provides a challenge for researchers, and we contribute to the literature concerning the identification of robust subgroups with clear biological interpretation. Specifically, we tackle this unsupervised problem via a novel rank‐based Bayesian clustering method. The advantages of our method are the integration and quantification of all uncertainties related to both the input data and the model, the probabilistic interpretation of final results to allow straightforward assessment of the stability of clusters leading to reliable conclusions, and the transparent biological interpretation of the identified clusters since each cluster is characterized by its top‐ranked genomic features. We applied our method to RNA‐seq data from cancer samples from 12 tumor types from the Cancer Genome Atlas. We identified a robust clustering that mostly reflects tissue of origin but also includes pan‐cancer clusters. Importantly, we identified three pan‐squamous clusters composed of a mix of lung squamous cell carcinoma, head and neck squamous carcinoma, and bladder cancer, with different biological functions over‐represented in the top genes that characterize the three clusters. We also found two novel subtypes of kidney cancer that show different prognosis, and we reproduced known subtypes of breast cancer. Taken together, our method allows the identification of robust and biologically meaningful clusters of pan‐cancer samples.

show abstract

BayesMallows: An R Package for the Bayesian Mallows Model

Cited by 10 publications

References 25 publications

Pseudo-Mallows for Efficient Probabilistic Preference Learning

Pseudo-Mallows for Efficient Probabilistic Preference Learning

Rank‐based Bayesian variable selection for genome‐wide transcriptomic analyses

Transcriptomic pan‐cancer analysis using rank‐based Bayesian inference

Contact Info

Product

Resources

About