Numerous biological and physical processes can be modeled as systems of interacting entities evolving continuously over time, e.g. the dynamics of communicating cells or physical particles. Learning the dynamics of such systems is essential for predicting the temporal evolution of populations across novel samples and unseen environments. Flow-based models allow for learning these dynamics at the population level - they model the evolution of the entire distribution of samples. However, current flow-based models are limited to a single initial population and a set of predefined conditions which describe different dynamics. We argue that multiple processes in natural sciences have to be represented as vector fields on the Wasserstein manifold of probability densities. That is, the change of the population at any moment in time depends on the population itself due to the interactions between samples. In particular, this is crucial for personalized medicine where the development of diseases and their respective treatment response depend on the microenvironment of cells specific to each patient. We propose *Meta Flow Matching* (MFM), a practical approach to integrate along these vector fields on the Wasserstein manifold by amortizing the flow model over the initial populations. Namely, we embed the population of samples using a Graph Neural Network (GNN) and use these embeddings to train a Flow Matching model. This gives MFM the ability to generalize over the initial distributions, unlike previously proposed methods. We demonstrate the ability of MFM to improve the prediction of individual treatment responses on a large-scale multi-patient single-cell drug screen dataset.

One of the grand challenges of cell biology is inferring the gene regulatory network (GRN) which describes interactions between genes and their products that control gene expression and cellular function. We can treat this as a causal discovery problem but with two non-standard challenges: (1) regulatory networks are inherently cyclic so we should not model a GRN as a directed acyclic graph (DAG), and (2) observations have significant measurement noise so for typical sample sizes, there will always be a large equivalence class of graphs that are likely given the data, and we want methods that capture this uncertainty. Existing methods either focus on challenge (1), identifying cyclic structure from dynamics, or on challenge (2) learning complex Bayesian posteriors over directed acyclic graphs, but not both. In this paper we leverage the fact that it is possible to estimate the ``velocity'' of the expression of a gene with RNA velocity techniques to develop an approach that addresses both challenges. Because we have access to velocity information, we can treat the Bayesian structure learning problem as a problem of sparse identification of a dynamical system, capturing cyclic feedback loops through time. We leverage Generative Flow Networks (GFlowNets) to estimate the posterior distribution over the combinatorial space of possible sparse dependencies. Our results indicate that our method learns posteriors that better encapsulate the distributions of cyclic structures compared to counterpart state-of-the-art Bayesian structure learning approaches.

MOTIVATION: Molecular carcinogenicity is a preventable cause of cancer, but systematically identifying carcinogenic compounds, which involves performing experiments on animal models, is expensive, time consuming and low throughput. As a result, carcinogenicity information is limited and building data-driven models with good prediction accuracy remains a major challenge. RESULTS: In this work, we propose CONCERTO, a deep learning model that uses a graph transformer in conjunction with a molecular fingerprint representation for carcinogenicity prediction from molecular structure. Special efforts have been made to overcome the data size constraint, such as multi-round pre-training on related but lower quality mutagenicity data, and transfer learning from a large self-supervised model. Extensive experiments demonstrate that our model performs well and can generalize to external validation sets. CONCERTO could be useful for guiding future carcinogenicity experiments and provide insight into the molecular basis of carcinogenicity. AVAILABILITY AND IMPLEMENTATION: The code and data underlying this article are available on github at https://github.com/bowang-lab/CONCERTO.

MOTIVATION: Alternative splicing (AS) is a regulated process that directs the generation of different transcripts from single genes. A computational model that can accurately predict splicing patterns based on genomic features and cellular context is highly desirable, both in understanding this widespread phenomenon, and in exploring the effects of genetic variations on AS. METHODS: Using a deep neural network, we developed a model inferred from mouse RNA-Seq data that can predict splicing patterns in individual tissues and differences in splicing patterns across tissues. Our architecture uses hidden variables that jointly represent features in genomic sequences and tissue types when making predictions. A graphics processing unit was used to greatly reduce the training time of our models with millions of parameters. RESULTS: We show that the deep architecture surpasses the performance of the previous Bayesian method for predicting AS patterns. With the proper optimization procedure and selection of hyperparameters, we demonstrate that deep architectures can be beneficial, even with a moderately sparse dataset. An analysis of what the model has learned in terms of the genomic features is presented.

Transcript quantification is a long-standing problem in genomics and estimating the relative abundance of alternatively-spliced isoforms from the same transcript is an important special case. Both problems have recently been illuminated by high-throughput RNA sequencing experiments which are quickly generating large amounts of data. However, much of the signal present in this data is corrupted or obscured by biases resulting in non-uniform and non-proportional representation of sequences from different transcripts. Many existing analyses attempt to deal with these and other biases with various task-specific approaches, which makes direct comparison between them difficult. However, two popular tools for isoform quantification, MISO and Cufflinks, have adopted a general probabilistic framework to model and mitigate these biases in a more general fashion. These advances motivate the need to investigate the effects of RNA-seq biases on the accuracy of different approaches for isoform quantification. We conduct the investigation by building models of increasing sophistication to account for noise introduced by the biases and compare their accuracy to the established approaches. We focus on methods that estimate the expression of alternatively-spliced isoforms with the percent-spliced-in (PSI) metric for each exon skipping event. To improve their estimates, many methods use evidence from RNA-seq reads that align to exon bodies. However, the methods we propose focus on reads that span only exon-exon junctions. As a result, our approaches are simpler and less sensitive to exon definitions than existing methods, which enables us to distinguish their strengths and weaknesses more easily. We present several probabilistic models of of position-specific read counts with increasing complexity and compare them to each other and to the current state-of-the-art methods in isoform quantification, MISO and Cufflinks. On a validation set with RT-PCR measurements for 26 cassette events, some of our methods are more accurate and some are significantly more consistent than these two popular tools. This comparison demonstrates the challenges in estimating the percent inclusion of alternatively spliced junctions and illuminates the tradeoffs between different approaches.

A novel Kalman filtering/smoothing algorithm is presented for efficient and accurate estimation of vocal tract resonances or formants, which are natural frequencies and bandwidths of the resonator from larynx to lips, in fluent speech. The algorithm uses a hidden dynamic model, with a state-space formulation, where the resonance frequency and bandwidth values are treated as continuous-valued hidden state variables. The observation equation of the model is constructed by an analytical predictive function from the resonance frequencies and bandwidths to LPC cepstra as the observation vectors. This nonlinear function is adaptively linearized, and a residual or bias term, which is adaptively trained, is added to the nonlinear function to represent the iteratively reduced piecewise linear approximation error. Details of the piecewise linearization design process are described. An iterative tracking algorithm is presented, which embeds both the adaptive residual training and piecewise linearization design in the Kalman filtering/smoothing framework. Experiments on estimating resonances in Switchboard speech data show accurate estimation results. In particular, the effectiveness of the adaptive residual training is demonstrated. Our approach provides a solution to the traditional "hidden formant problem," and produces meaningful results even during consonantal closures when the supra-laryngeal source may cause no spectral prominences in speech acoustics

A better understanding of cell behavior is very important in drug and disease research. Cell size, shape, and motility may play a key role in stem-cell specialization or cancer development. However the traditional method of inferring these values from image sequences manually is such an onerous task that automated methods of cell tracking and segmentation are in high demanded, especially given the increasing amount of cell data being collected. In this paper, a novel probabilistic cell model is designed to segment the individual hematopoietic stem cells (HSCs) extracted from mice bone marrow cells. The proposed cell model has been successfully applied to HSC segmentation, identifying the most probable cell locations in the image on the basis of cell brightness and morphology.

A novel approach is developed for efficient and accurate tracking of vocal tract resonances, which are natural frequencies of the resonator from larynx to lips, in fluent speech. The tracking algorithm is based on a version of the structured speech model consisting of continuous-valued hidden dynamics and a piecewise-linearized prediction function from resonance frequencies and bandwidths to LPC cepstra. We present details of the piecewise linearization design process and an adaptive training technique for the parameters that characterize the prediction residuals. An iterative tracking algorithm is described and evaluated that embeds both the prediction-residual training and the piecewise linearization design in an adaptive Kalman filtering framework. Experiments on tracking vocal tract resonances in Switchboard speech data demonstrate high accuracy in the results, as well as the effectiveness of residual training embedded in the algorithm. Our approach differs from traditional formant trackers in that it provides meaningful results even during consonantal closures when the supra-laryngeal source may cause no spectral prominences in speech acoustics.

An important general model for discrete-time signal processing is the switching state space (SSS) model, which generalizes the hidden Markov model and the Gaussian state space model. Inference and parameter estimation in this model are known to be computationally intractable. This paper presents a powerful new approximation to the SSS model. The approximation is based on a variational technique that preserves the multimodal nature of the continuous state posterior distribution. Furthermore, by incorporating a windowing technique, the resulting EM algorithm has complexity that is just linear in the length of the time series. An alternative Viterbi decoding with frame-based likelihood is also presented which is crucial for the speech application that originally motivates this work. Our experiments focus on demonstrating the effectiveness of the algorithm by extensive simulations. A typical example in speech processing is also included to show the potential of this approach for practical applications.

This paper describes novel and powerful variational EM algorithms for the segmental switching state space models used in speech applications, which are capable of capturing key internal (or hidden) dynamics of natural speech production. Hidden dynamic models (HDMs) have recently become a class of promising acoustic models to incorporate crucial speech-specific knowledge and overcome many inherent weaknesses of traditional HMMs. However, the lack of powerful and efficient statistical learning algorithms is one of the main obstacles preventing them from being well studied and widely used. Since exact inference and learning are intractable, a variational approach is taken to develop effective approximate algorithms. We have implemented the segmental constraint crucial for modeling speech dynamics and present algorithms for recovering hidden speech dynamics and discrete speech units from acoustic data only. The effectiveness of the algorithms developed are verified by experiments on simulation and Switchboard speech data.

Virtually all implementations of simulated annealing are simplified by assuming discrete unknowns, however continuous-parameter annealing has many potential applications to image processing. Widely scattered problems such as formant tracking, boundary estimation and phase- unwrapping can all be approached as the annealed minimizations of continuous B-spline parameters. The benefits of simulated annealing are well known, including an insensitivity to initial conditions and the ability to solve problems with many local minima. Discrete variable annealing has seen broad application, however continuous-variable annealing is limited by the computational challenge of Gibbs sampling. In this paper we develop efficient approaches to sampling, illustrated in the context of contour tracking in noisy images.

Introduces a statistical speech production model. The model synthesizes natural speech by modeling some key dynamic properties of vocal articulators in a linear/nonlinear state-space framework. The goal-oriented movements of the articulators (tongue tip, tongue dorsum, upper lip, lower lip, and jaw) are described in a linear dynamic state equation. The resulting articulatory trajectories, combined with the effects of the velum and larynx, are nonlinearly mapped into the acoustic feature space (MFCCs). The key challenges in this model are the development of a nonlinear parameter estimation methodology, and the incorporation of appropriate prior assumptions to assert in the articulatory dynamic structure. Such a model can also be directly applied to speech recognition to better account for coarticulation and phonetic reduction phenomena with considerably fewer parameters than HMM based approaches.