Ensemble methods, particularly boosting, have established themselves as highly effective and widely embraced machine learning techniques for tabular data. In this paper, we aim to leverage the robust predictive power of traditional boosting methods while enhancing fairness and interpretability. To achieve this, we develop FAIR MP-BOOST, a stochastic boosting scheme that balances fairness and accuracy by adaptively learning features and observations during training. Specifically, FAIR MP-BOOST sequentially samples small subsets of observations and features, termed minipatches (MP), according to adaptively learned feature and observation sampling probabilities. We devise these probabilities by combining loss functions, or by combining feature importance scores to address accuracy and fairness simultaneously. Hence, FAIR MP-BOOST prioritizes important and fair features along with challenging instances, to select the most relevant minipatches for learning. The learned probability distributions also yield intrinsic interpretations of feature importance and important observations in FAIR MP-BOOST. Through empirical evaluation of simulated and benchmark datasets, we showcase the interpretability, accuracy, and fairness of FAIR MP-BOOST.

Characterizing interactions between brain areas is a fundamental goal of systems neuroscience. While such analyses are possible when areas are recorded simultaneously, it is rare to observe all combinations of areas of interest within a single animal or recording session. How can we leverage multi-animal datasets to better understand multi-area interactions? Building on recent progress in large-scale, multi-animal models, we introduce NeuroPaint, a masked autoencoding approach for inferring the dynamics of unobserved brain areas. By training across animals with overlapping subsets of recorded areas, NeuroPaint learns to reconstruct activity in missing areas based on shared structure across individuals. We train and evaluate our approach on both synthetic data and two multi-animal, multi-area Neuropixels datasets. Our results demonstrate that models trained across animals with partial observations can successfully in-paint the dynamics of unrecorded areas, enabling multi-area analyses that transcend the limitations of any single experiment.

In this Innovative-Practice Full Paper, we present the design and implementation of a novel strategy for the assessment of the individual contributions of students working in teams for an interdisciplinary project-based learning (PBL) course. Team-based and project-based learning have been the cornerstones of experiential learning in recent years. In such courses, the assessment of students is often performed solely at a team level, resulting in student discontent, concerns about fairness, and social loafing. The application of a single method for evaluating individual contributions has its shortcomings, prompting us to incorporate a combination of evidence-based approaches. Thus, to mitigate these problems, we developed an assessment strategy that involves three components - self and peer evaluation (SPA), instructor's evaluation of individual contribution, and class participation. Anonymized SPA was carried out at three different checkpoints over the semester for both formative and summative assessment. The individual contribution of the students is also evaluated by the instructor and/or other mentors based on their interactions with all the students in the team over the entire semester. Class participation is another component of individual contribution incorporated in our scheme, where instructors evaluate the participation of individual students in classroom activities, presentations, and meetings. We collected feedback about the perception of students towards our assessment policies for individual contribution. We observed overall satisfaction and positive attitudes toward our scheme. We noted positive student perceptions of fairness in grading through our scheme and reduced chances of social loafing. Further, in the feedback, the students noted the effectiveness of using SPA as an evaluation tool, the usefulness of instructor's evaluation, and the role of class participation in creating a more engaging classroom and a more enriching experience. We also emphasize the need to set clear expectations for students with regard to the grading policies early in the semester. Finally, we discuss our experiences in implementing our design and recommendations for incorporating our grading scheme in team-based PBL courses.

In this work, we propose data augmentation via pairwise mixup across subgroups to improve group fairness. Many real-world applications of machine learning systems exhibit biases across certain groups due to underrepresentation or training data that reflects societal biases. Inspired by the successes of mixup for improving classification performance, we develop a pairwise mixup scheme to augment training data and encourage fair and accurate decision boundaries for all subgroups. Data augmentation for group fairness allows us to add new samples of underrepresented groups to balance subpopulations. Furthermore, our method allows us to use the generalization ability of mixup to improve both fairness and accuracy. We compare our proposed mixup to existing data augmentation and bias mitigation approaches on both synthetic simulations and real-world benchmark fair classification data, demonstrating that we are able to achieve fair outcomes with robust if not improved accuracy.

Elucidating functional connectivity reveals interactions between brain regions based on their in vivo activity. However, a gap exists in current statistical methods’ ability to untangle direct interactions from confounding activity due to stimuli and physiological effects, leading to a bottleneck in understanding how the brain filters and decodes information. We propose methods to remove physiological and environmental effects from calcium imaging in the rodent olfactory bulb to better isolate connectivity-related activity correlations. Our approach deconvolves brain signals by learning smooth filters leveraging generalized ridge regularization, thus producing stronger isolation of such connectivity-related activity. Our method for removing specific external correlation-causing factors is applicable in a wide variety of settings including those with different external stimuli.

MOTIVATION: Nuclear magnetic resonance spectroscopy (NMR) is widely used to analyze metabolites in biological samples, but the analysis requires specific expertise, it is time-consuming, and can be inaccurate. Here, we present a powerful automate tool, SPatial clustering Algorithm-Statistical TOtal Correlation SpectroscopY (SPA-STOCSY), which overcomes challenges faced when analyzing NMR data and identifies metabolites in a sample with high accuracy. RESULTS: As a data-driven method, SPA-STOCSY estimates all parameters from the input dataset. It first investigates the covariance pattern among datapoints and then calculates the optimal threshold with which to cluster datapoints belonging to the same structural unit, i.e. the metabolite. Generated clusters are then automatically linked to a metabolite library to identify candidates. To assess SPA-STOCSY's efficiency and accuracy, we applied it to synthesized spectra and spectra acquired on Drosophila melanogaster tissue and human embryonic stem cells. In the synthesized spectra, SPA outperformed Statistical Recoupling of Variables (SRV), an existing method for clustering spectral peaks, by capturing a higher percentage of the signal regions and the close-to-zero noise regions. In the biological data, SPA-STOCSY performed comparably to the operator-based Chenomx analysis while avoiding operator bias, and it required <7 min of total computation time. Overall, SPA-STOCSY is a fast, accurate, and unbiased tool for untargeted analysis of metabolites in the NMR spectra. It may thus accelerate the use of NMR for scientific discoveries, medical diagnostics, and patient-specific decision making. AVAILABILITY AND IMPLEMENTATION: The codes of SPA-STOCSY are available at https://github.com/LiuzLab/SPA-STOCSY.

This Innovative-Practice Full Paper presents the curriculum development and our experiences in offering a client-facing consulting course in data science. Data science education has seen rapid growth over the past decade. To provide students with hands-on opportunities to work with real data, many data science programs have advocated for and implemented experiential learning opportunities throughout the curriculum, which has been shown in a wide variety of literature to have many benefits. Most experiential learning opportunities in STEM programs are provided through capstone and engineering design courses; this is becoming increasingly the case in data science programs as well where several universities have developed data science capstone programs in which students work with clients on the client’s real-world data sets. While client-sponsored capstone projects are an exemplar of experiential learning, they may pose major challenges to implement and can be particularly resource-intensive for institutions; this is especially the case in data science where the legalities of data sharing may come with additional hurdles. Because of this, we were motivated to develop a novel client-facing data science consulting course that provides a unique experiential learning scenario to both undergraduate and graduate students while requiring much fewer resources and legalities. In our novel data science consulting course, groups of students work directly with real clients in a consulting clinic setting to provide data science guidance and short-term help with data science challenges. Through this process, students learn about the diversity of real-world problems in data science, how to lead consultations with clients effectively as a team, how to frame data science challenges and research possible solutions, and how to communicate solutions to clients in reports and presentations. We leveraged best practices in consulting courses developed in business school settings to design our course. Additionally, the consulting course serves as a community service initiative whereby researchers, clinicians, non-profit and government workers, and industry professionals benefit from the advice and short-term help provided through consultation. In this paper, we report how our consulting course is set up, how clients from both within and outside the university can seek help at the consulting clinic, and how the structure of the course enables students to have firsthand experience working on many real-world data science problems with clients. Finally, we discuss how student performance is assessed in this course, the lessons learned from offering this course, and recommendations for other data science programs in universities that wish to design similar courses.

High-dimensional Gaussian graphical models have been a powerful tool for learning connections or interaction patterns among a large number of variables. While most prior work focuses on the case when all variables are measured simultaneously, we consider a practically common but new setting when no simultaneous measurement of all variables is available and there are differing numbers of measurements on each pair of nodes. This occurs in estimating gene expression networks from single-cell sequencing data, functional connectivity from neuronal recordings, and sensor networks. Incorporating a projection step in the estimation of the covariance matrix, we develop a convex method for learning the neighborhood of any node. We provide theoretical guarantees for the proposed method, suggesting that the exact recovery of the neighborhood of any node a hinges on sufficient joint measurements for the pairs of nodes that involve a’s neighbor(s). We also present some consequences of the theory and numerical experiments on special graphs.

Consensus clustering has been widely used in bioinformatics and other applications to improve the accuracy, stability and reliability of clustering results. This approach ensembles cluster co-occurrences from multiple clustering runs on subsampled observations. For application to large-scale bioinformatics data, such as to discover cell types from single-cell sequencing data, for example, consensus clustering has two significant drawbacks: (i) computational inefficiency due to repeatedly applying clustering algorithms, and (ii) lack of interpretability into the important features for differentiating clusters. In this paper, we address these two challenges by developing IMPACC: Interpretable MiniPatch Adaptive Consensus Clustering. Our approach adopts three major innovations. We ensemble cluster co-occurrences from tiny subsets of both observations and features, termed minipatches, thus dramatically reducing computation time. Additionally, we develop adaptive sampling schemes for observations, which result in both improved reliability and computational savings, as well as adaptive sampling schemes of features, which lead to interpretable solutions by quickly learning the most relevant features that differentiate clusters. We study our approach on synthetic data and a variety of real large-scale bioinformatics data sets; results show that our approach not only yields more accurate and interpretable cluster solutions, but it also substantially improves computational efficiency compared to standard consensus clustering approaches.

Interest in data science education and degree programs has rapidly expanded over the past several years. An integral part of many degree programs is a capstone experience, where students complete a major research or real-world project at the culmination of their educational program. In engineering and computer science, many have shown that client-sponsored projects lead to better student engagement and improved training. In this paper, we discuss experiences with developing an interdisciplinary, client-sponsored capstone program in data science and machine learning. We show how we set up the capstone program, including how the program is structured, how projects are set up, how the course is managed, how students are assessed, and outline the newly developed capstone curriculum. Finally, we report results from a cohort of students participating in this capstone program and discuss lessons learned as well as best practices when developing data science capstone programs.

Data integration, or the strategic analysis of multiple sources of data simultaneously, can often lead to discoveries that may be hidden in individualistic analyses of a single data source. We develop a new unsupervised data integration method named Integrated Principal Components Analysis (iPCA), which is a model-based generalization of PCA and serves as a practical tool to find and visualize common patterns that occur in multiple data sets. The key idea driving iPCA is the matrix-variate normal model, whose Kronecker product covariance structure captures both individual patterns within each data set and joint patterns shared by multiple data sets. Building upon this model, we develop several penalized (sparse and non-sparse) covariance estimators for iPCA, and using geodesic convexity, we prove that our non-sparse iPCA estimator converges to the global solution of a non-convex problem. We also demonstrate the practical advantages of iPCA through extensive simulations and a case study application to integrative genomics for Alzheimer's disease. In particular, we show that the joint patterns extracted via iPCA are highly predictive of a patient's cognition and Alzheimer's diagnosis.

In mixed multi-view data, multiple sets of diverse features are measured on the same set of samples. By integrating all available data sources, we seek to discover common group structure among the samples that may be hidden in individualistic cluster analyses of a single data view. While several techniques for such integrative clustering have been explored, we propose and develop a convex formalization that enjoys strong empirical performance and inherits the mathematical properties of increasingly popular convex clustering methods. Specifically, our Integrative Generalized Convex Clustering Optimization (iGecco) method employs different convex distances, losses, or divergences for each of the different data views with a joint convex fusion penalty that leads to common groups. Additionally, integrating mixed multi-view data is often challenging when each data source is high-dimensional. To perform feature selection in such scenarios, we develop an adaptive shifted group-lasso penalty that selects features by shrinking them towards their loss-specific centers. Our so-called iGecco+ approach selects features from each data view that are best for determining the groups, often leading to improved integrative clustering. To solve our problem, we develop a new type of generalized multi-block ADMM algorithm using sub-problem approximations that more efficiently fits our model for big data sets. Through a series of numerical experiments and real data examples on text mining and genomics, we show that iGecco+ achieves superior empirical performance for high-dimensional mixed multi-view data.

ElectroCOrticoGraphy (ECoG) technology measures electrical activity in the human brain via electrodes placed directly on the cortical surface during neurosurgery. Through its capability to record activity at an extremely fast temporal resolution, ECoG experiments have allowed scientists to better understand how the human brain processes speech. By its nature, ECoG data is extremely difficult for neuroscientists to directly interpret for two major reasons. Firstly, ECoG data tends to be extremely large in size, as each individual experiment yields data up to several GB. Secondly, ECoG data has a complex, higher-order nature; after signal processing, this type of data is typically organized as a 4-way tensor consisting of trials by electrodes by frequency by time. In this paper, we develop an interpretable dimension reduction approach called Regularized Higher Order Principal Components Analysis, as well as an extension to Regularized Higher Order Partial Least Squares, that allows neuroscientists to explore and visualize ECoG data. Our approach employs a sparse and functional Candecomp-Parafac (CP) decomposition that incorporates sparsity to select relevant electrodes and frequency bands, as well as smoothness over time and frequency, yielding directly interpretable factors. We demonstrate both the performance and interpretability of our method with an ECoG case study on audio and visual processing of human speech.

Motivation: Batch effects are one of the major source of technical variations that affect the measurements in high-throughput studies such as RNA sequencing. It has been well established that batch effects can be caused by different experimental platforms, laboratory conditions, different sources of samples and personnel differences. These differences can confound the outcomes of interest and lead to spurious results. A critical input for batch correction algorithms is the knowledge of batch factors, which in many cases are unknown or inaccurate. Hence, the primary motivation of our paper is to detect hidden batch factors that can be used in standard techniques to accurately capture the relationship between gene expression and other modeled variables of interest. Results: We introduce a new algorithm based on data-adaptive shrinkage and semi-Non-negative Matrix Factorization for the detection of unknown batch effects. We test our algorithm on three different datasets: (i) Sequencing Quality Control, (ii) Topotecan RNA-Seq and (iii) Single-cell RNA sequencing (scRNA-Seq) on Glioblastoma Multiforme. We have demonstrated a superior performance in identifying hidden batch effects as compared to existing algorithms for batch detection in all three datasets. In the Topotecan study, we were able to identify a new batch factor that has been missed by the original study, leading to under-representation of differentially expressed genes. For scRNA-Seq, we demonstrated the power of our method in detecting subtle batch effects. Availability and implementation: DASC R package is available via Bioconductor or at https://github.com/zhanglabNKU/DASC. Contact: [email protected] or [email protected]. Supplementary information: Supplementary data are available at Bioinformatics online.

The 2016 International Conference on Intelligent Biology and Medicine (ICIBM 2016) was held on December 8-10, 2016 in Houston, Texas, USA. ICIBM included eight scientific sessions, four tutorials, one poster session, four highlighted talks and four keynotes that covered topics on 3D genomics structural analysis, next generation sequencing (NGS) analysis, computational drug discovery, medical informatics, cancer genomics, and systems biology. Here, we present a summary of the nine research articles selected from ICIBM 2016 program for publishing in BMC Bioinformatics.

MOTIVATION: Massive amounts of high-throughput genomics data profiled from tumor samples were made publicly available by the Cancer Genome Atlas (TCGA). RESULTS: We have developed an open source software package, TCGA2STAT, to obtain the TCGA data, wrangle it, and pre-process it into a format ready for multivariate and integrated statistical analysis in the R environment. In a user-friendly format with one single function call, our package downloads and fully processes the desired TCGA data to be seamlessly integrated into a computational analysis pipeline. No further technical or biological knowledge is needed to utilize our software, thus making TCGA data easily accessible to data scientists without specific domain knowledge. AVAILABILITY AND IMPLEMENTATION: TCGA2STAT is available from the https://cran.r-project.org/web/packages/TCGA2STAT/index.html SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online. CONTACT: [email protected].

Markov Random Fields, or undirected graphical models are widely used to model high-dimensional multivariate data. Classical instances of these models, such as Gaussian Graphical and Ising Models, as well as recent extensions to graphical models specified by univariate exponential families, assume all variables arise from the same distribution. Complex data from high-throughput genomics and social networking for example, often contain discrete, count, and continuous variables measured on the same set of samples. To model such heterogeneous data, we develop a \emphnovel class of mixed graphical models by specifying that each node-conditional distribution is a member of a possibly different univariate exponential family. We study several instances of our model, and propose scalable M-estimators for recovering the underlying network structure. Simulations as well as an application to learning mixed genomic networks from next generation sequencing and mutation data demonstrate the versatility of our methods.

Conditional random fields, which model the distribution of a multivariate response conditioned on a set of covariates using undirected graphs, are widely used in a variety of multivariate prediction applications. Popular instances of this class of models such as categorical-discrete CRFs, Ising CRFs, and conditional Gaussian based CRFs, are not however best suited to the varied types of response variables in many applications, including count-valued responses. We thus introduce a “novel subclass of CRFs”, derived by imposing node-wise conditional distributions of response variables conditioned on the rest of the responses and the covariates as arising from univariate exponential families. This allows us to derive novel multivariate CRFs given any univariate exponential distribution, including the Poisson, negative binomial, and exponential distributions. Also in particular, it addresses the common CRF problem of specifying feature'' functions determining the interactions between response variables and covariates. We develop a class of tractable penalized $M$-estimators to learn these CRF distributions from data, as well as a unified sparsistency analysis for this general class of CRFs showing exact structure recovery can be achieved with high probability."

Undirected graphical models, such as Gaussian graphical models, Ising, and multinomial/categorical graphical models, are widely used in a variety of applications for modeling distributions over a large number of variables. These standard instances, however, are ill-suited to modeling count data, which are increasingly ubiquitous in big-data settings such as genomic sequencing data, user-ratings data, spatial incidence data, climate studies, and site visits. Existing classes of Poisson graphical models, which arise as the joint distributions that correspond to Poisson distributed node-conditional distributions, have a major drawback: they can only model negative conditional dependencies for reasons of normalizability given its infinite domain. In this paper, our objective is to modify the Poisson graphical model distribution so that it can capture a rich dependence structure between count-valued variables. We begin by discussing two strategies for truncating the Poisson distribution and show that only one of these leads to a valid joint distribution; even this model, however, has limitations on the types of variables and dependencies that may be modeled. To address this, we propose two novel variants of the Poisson distribution and their corresponding joint graphical model distributions. These models provide a class of Poisson graphical models that can capture both positive and negative conditional dependencies between count-valued variables. One can learn the graph structure of our model via penalized neighborhood selection, and we demonstrate the performance of our methods by learning simulated networks as well as a network from microRNA-Sequencing data.

Gaussian graphical models are often used to infer gene networks based on microarray expression data. Many scientists, however, have begun using high-throughput sequencing technologies to measure gene expression. As the resulting high-dimensional count data consists of counts of sequencing reads for each gene, Gaussian graphical models are not optimal for modeling gene networks based on this discrete data. We develop a novel method for estimating high-dimensional Poisson graphical models, the Log-Linear Graphical Model, allowing us to infer networks based on high-throughput sequencing data. Our model assumes a pair-wise Markov property: conditional on all other variables, each variable is Poisson. We estimate our model locally via neighborhood selection by fitting 1-norm penalized log-linear models. Additionally, we develop a fast parallel algorithm permitting us to fit our graphical model to high-dimensional genomic data sets. We illustrate the effectiveness of our methods for recovering network structure from count data through simulations and a case study on breast cancer microRNA networks.

Undirected graphical models, or Markov networks, such as Gaussian graphical models and Ising models enjoy popularity in a variety of applications. In many settings, however, data may not follow a Gaussian or binomial distribution assumed by these models. We introduce a new class of graphical models based on generalized linear models (GLM) by assuming that node-wise conditional distributions arise from exponential families. Our models allow one to estimate networks for a wide class of exponential distributions, such as the Poisson, negative binomial, and exponential, by fitting penalized GLMs to select the neighborhood for each node. A major contribution of this paper is the rigorous statistical analysis showing that with high probability, the neighborhood of our graphical models can be recovered exactly. We provide examples of high-throughput genomic networks learned via our GLM graphical models for multinomial and Poisson distributed data.

MOTIVATION: Nuclear magnetic resonance (NMR) spectroscopy has been used to study mixtures of metabolites in biological samples. This technology produces a spectrum for each sample depicting the chemical shifts at which an unknown number of latent metabolites resonate. The interpretation of this data with common multivariate exploratory methods such as principal components analysis (PCA) is limited due to high-dimensionality, non-negativity of the underlying spectra and dependencies at adjacent chemical shifts. RESULTS: We develop a novel modification of PCA that is appropriate for analysis of NMR data, entitled Sparse Non-Negative Generalized PCA. This method yields interpretable principal components and loading vectors that select important features and directly account for both the non-negativity of the underlying spectra and dependencies at adjacent chemical shifts. Through the reanalysis of experimental NMR data on five purified neural cell types, we demonstrate the utility of our methods for dimension reduction, pattern recognition, sample exploration and feature selection. Our methods lead to the identification of novel metabolites that reflect the differences between these cell types. AVAILABILITY: www.stat.rice.edu/~gallen/software.html. CONTACT: [email protected]. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.