Reference-free inferring of transcriptomic events in cancer cells on single-cell data.

BACKGROUND: Cancerous cells' identity is determined via a mixture of multiple factors such as genomic variations, epigenetics, and the regulatory variations that are involved in transcription. The differences in transcriptome expression as well as abnormal structures in peptides determine phenotypical differences. Thus, bulk RNA-seq and more recent single-cell RNA-seq data (scRNA-seq) are important to identify pathogenic differences. In this case, we rely on k-mer decomposition of sequences to identify pathogenic variations in detail which does not need a reference, so it outperforms more traditional Next-Generation Sequencing (NGS) analysis techniques depending on the alignment of the sequences to a reference. RESULTS: Via our alignment-free analysis, over esophageal and glioblastoma cancer patients, high-frequency variations over multiple different locations (repeats, intergenic regions, exons, introns) as well as multiple different forms (fusion, polyadenylation, splicing, etc.) could be discovered. Additionally, we have analyzed the importance of less-focused events systematically in a classic transcriptome analysis pipeline where these events are considered as indicators for tumor prognosis, tumor prediction, tumor neoantigen inference, as well as their connection with respect to the immune microenvironment. CONCLUSIONS: Our results suggest that esophageal cancer (ESCA) and glioblastoma processes can be explained via pathogenic microbial RNA, repeated sequences, novel splicing variants, and long intergenic non-coding RNAs (lincRNAs). We expect our application of reference-free process and analysis to be helpful in tumor and normal samples differential scRNA-seq analysis, which in turn offers a more comprehensive scheme for major cancer-associated events.

BERT2OME: Prediction of 2'-O-Methylation Modifications From RNA Sequence by Transformer Architecture Based on BERT.

Recent work on language models has resulted in state-of-the-art performance on various language tasks. Among these, Bidirectional Encoder Representations from Transformers (BERT) has focused on contextualizing word embeddings to extract context and semantics of the words. On the other hand, post-transcriptional 2'-O-methylation (Nm) RNA modification is important in various cellular tasks and related to a number of diseases. The existing high-throughput experimental techniques take longer time to detect these modifications, and costly in exploring these functional processes. Here, to deeply understand the associated biological processes faster, we come up with an efficient method Bert2Ome to infer 2'-O-methylation RNA modification sites from RNA sequences. Bert2Ome combines BERT-based model with convolutional neural networks (CNN) to infer the relationship between the modification sites and RNA sequence content. Unlike the methods proposed so far, Bert2Ome assumes each given RNA sequence as a text and focuses on improving the modification prediction performance by integrating the pretrained deep learning-based language model BERT. Additionally, our transformer-based approach could infer modification sites across multiple species. According to 5-fold cross-validation, human and mouse accuracies were 99.15% and 94.35% respectively. Similarly, ROC AUC scores were 0.99, 0.94 for the same species. Detailed results show that Bert2Ome reduces the time consumed in biological experiments and outperforms the existing approaches across different datasets and species over multiple metrics. Additionally, deep learning approaches such as 2D CNNs are more promising in learning BERT attributes than more conventional machine learning methods.

A comparison of topologically associating domain callers over mammals at high resolution.

BACKGROUND: Topologically associating domains (TADs) are locally highly-interacting genome regions, which also play a critical role in regulating gene expression in the cell. TADs have been first identified while investigating the 3D genome structure over High-throughput Chromosome Conformation Capture (Hi-C) interaction dataset. Substantial degree of efforts have been devoted to develop techniques for inferring TADs from Hi-C interaction dataset. Many TAD-calling methods have been developed which differ in their criteria and assumptions in TAD inference. Correspondingly, TADs inferred via these callers vary in terms of both similarities and biological features they are enriched in. RESULT: We have carried out a systematic comparison of 27 TAD-calling methods over mammals. We use Micro-C, a recent high-resolution variant of Hi-C, to compare TADs at a very high resolution, and classify the methods into 3 categories: feature-based methods, Clustering methods, Graph-partitioning methods. We have evaluated TAD boundaries, gaps between adjacent TADs, and quality of TADs across various criteria. We also found particularly CTCF and Cohesin proteins to be effective in formation of TADs with corner dots. We have also assessed the callers performance on simulated datasets since a gold standard for TADs is missing. TAD sizes and numbers change remarkably between TAD callers and dataset resolutions, indicating that TADs are hierarchically-organized domains, instead of disjoint regions. A core subset of feature-based TAD callers regularly perform the best while inferring reproducible domains, which are also enriched for TAD related biological properties. CONCLUSION: We have analyzed the fundamental principles of TAD-calling methods, and identified the existing situation in TAD inference across high resolution Micro-C interaction datasets over mammals. We come up with a systematic, comprehensive, and concise framework to evaluate the TAD-calling methods performance across Micro-C datasets. Our research will be useful in selecting appropriate methods for TAD inference and evaluation based on available data, experimental design, and biological question of interest. We also introduce our analysis as a benchmarking tool with publicly available source code.

Biocode: A Data-Driven Procedure to Learn the Growth of Biological Networks.

Probabilistic biological network growth models have been utilized for many tasks including but not limited to capturing mechanism and dynamics of biological growth activities, null model representation, capturing anomalies, etc. Well-known examples of these probabilistic models are Kronecker model, preferential attachment model, and duplication-based model. However, we should frequently keep developing new models to better fit and explain the observed network features while new networks are being observed. Additionally, it is difficult to develop a growth model each time we study a new network. In this paper, we propose Biocode, a framework to automatically discover novel biological growth models matching user-specified graph attributes in directed and undirected biological graphs. Biocode designs a basic set of instructions which are common enough to model a number of well-known biological graph growth models. We combine such instruction-wise representation with a genetic algorithm based optimization procedure to encode models for various biological networks. We mainly evaluate the performance of Biocode in discovering models for biological collaboration networks, gene regulatory networks, and protein interaction networks which features such as assortativity, clustering coefficient, degree distribution closely match with the true ones in the corresponding real biological networks.

ProbC: joint modeling of epigenome and transcriptome effects in 3D genome.

BACKGROUND: Hi-C and its high nucleosome resolution variant Micro-C provide a window into the spatial packing of a genome in 3D within the cell. Even though both techniques do not directly depend on the binding of specific antibodies, previous work has revealed enriched interactions and domain structures around multiple chromatin marks; epigenetic modifications and transcription factor binding sites. However, the joint impact of chromatin marks in Hi-C and Micro-C interactions have not been globally characterized, which limits our understanding of 3D genome characteristics. An emerging question is whether it is possible to deduce 3D genome characteristics and interactions by integrative analysis of multiple chromatin marks and associate interactions to functionality of the interacting loci. RESULT: We come up with a probabilistic method PROBC to decompose Hi-C and Micro-C interactions by known chromatin marks. PROBC is based on convex likelihood optimization, which can directly take into account both interaction existence and nonexistence. Through PROBC, we discover histone modifications (H3K27ac, H3K9me3, H3K4me3, H3K4me1) and CTCF as particularly predictive of Hi-C and Micro-C contacts across cell types and species. Moreover, histone modifications are more effective than transcription factor binding sites in explaining the genome's 3D shape through these interactions. PROBC can successfully predict Hi-C and Micro-C interactions in given species, while it is trained on different cell types or species. For instance, it can predict missing nucleosome resolution Micro-C interactions in human ES cells trained on mouse ES cells only from these 5 chromatin marks with above 0.75 AUC. Additionally, PROBC outperforms the existing methods in predicting interactions across almost all chromosomes. CONCLUSION: Via our proposed method, we optimally decompose Hi-C interactions in terms of these chromatin marks at genome and chromosome levels. We find a subset of histone modifications and transcription factor binding sites to be predictive of both Hi-C and Micro-C interactions and TADs across human, mouse, and different cell types. Through learned models, we can predict interactions on species just from chromatin marks for which Hi-C data may be limited.

Metric Labeling and Semimetric Embedding for Protein Annotation Prediction.

Computational techniques have been successful at predicting protein function from relational data (functional or physical interactions). These techniques have been used to generate hypotheses and to direct experimental validation. With few exceptions, the task is modeled as multilabel classification problems where the labels (functions) are treated independently or semi-independently. However, databases such as the Gene Ontology provide information about the similarities between functions. We explore the use of the Metric Labeling combinatorial optimization problem to make use of heuristically computed distances between functions to make more accurate predictions of protein function in networks derived from both physical interactions and a combination of other data types. To do this, we give a new technique (based on convex optimization) for converting heuristic semimetric distances into a metric with minimum least-squared distortion (LSD). The Metric Labeling approach is shown to outperform five existing techniques for inferring function from networks. These results suggest that Metric Labeling is useful for protein function prediction, and that LSD minimization can help solve the problem of converting heuristic distances to a metric.

Semi-nonparametric modeling of topological domain formation from epigenetic data.

BACKGROUND: Hi-C experiments capturing the 3D genome architecture have led to the discovery of topologically-associated domains (TADs) that form an important part of the 3D genome organization and appear to play a role in gene regulation and other functions. Several histone modifications have been independently associated with TAD formation, but their combinatorial effects on domain formation remain poorly understood at a global scale. RESULTS: We propose a convex semi-nonparametric approach called nTDP based on Bernstein polynomials to explore the joint effects of histone markers on TAD formation as well as predict TADs solely from the histone data. We find a small subset of modifications to be predictive of TADs across species. By inferring TADs using our trained model, we are able to predict TADs across different species and cell types, without the use of Hi-C data, suggesting their effect is conserved. This work provides the first comprehensive joint model of the effect of histone markers on domain formation. CONCLUSIONS: Our approach, nTDP, can form the basis of a unified, explanatory model of the relationship between epigenetic marks and topological domain structures. It can be used to predict domain boundaries for cell types, species, and conditions for which no Hi-C data is available. The model may also be of use for improving Hi-C-based domain finders.

Selecting the most appropriate time points to profile in high-throughput studies.

Biological systems are increasingly being studied by high throughput profiling of molecular data over time. Determining the set of time points to sample in studies that profile several different types of molecular data is still challenging. Here we present the Time Point Selection (TPS) method that solves this combinatorial problem in a principled and practical way. TPS utilizes expression data from a small set of genes sampled at a high rate. As we show by applying TPS to study mouse lung development, the points selected by TPS can be used to reconstruct an accurate representation for the expression values of the non selected points. Further, even though the selection is only based on gene expression, these points are also appropriate for representing a much larger set of protein, miRNA and DNA methylation changes over time. TPS can thus serve as a key design strategy for high throughput time series experiments. Supporting Website: www.sb.cs.cmu.edu/TPS.

Diffusion archeology for diffusion progression history reconstruction.

Diffusion through graphs can be used to model many real-world processes, such as the spread of diseases, social network memes, computer viruses, or water contaminants. Often, a real-world diffusion cannot be directly observed while it is occurring - perhaps it is not noticed until some time has passed, continuous monitoring is too costly, or privacy concerns limit data access. This leads to the need to reconstruct how the present state of the diffusion came to be from partial diffusion data. Here, we tackle the problem of reconstructing a diffusion history from one or more snapshots of the diffusion state. This ability can be invaluable to learn when certain computer nodes are infected or which people are the initial disease spreaders to control future diffusions. We formulate this problem over discrete-time SEIRS-type diffusion models in terms of maximum likelihood. We design methods that are based on submodularity and a novel prize-collecting dominating-set vertex cover (PCDSVC) relaxation that can identify likely diffusion steps with some provable performance guarantees. Our methods are the first to be able to reconstruct complete diffusion histories accurately in real and simulated situations. As a special case, they can also identify the initial spreaders better than the existing methods for that problem. Our results for both meme and contaminant diffusion show that the partial diffusion data problem can be overcome with proper modeling and methods, and that hidden temporal characteristics of diffusion can be predicted from limited data.

Tradeoffs between Dense and Replicate Sampling Strategies for High-Throughput Time Series Experiments.

An important experimental design question for high-throughput time series studies is the number of replicates required for accurate reconstruction of the profiles. Due to budget and sample availability constraints, more replicates imply fewer time points and vice versa. We analyze the performance of dense and replicate sampling by developing a theoretical framework that focuses on a restricted yet expressive set of possible curves over a wide range of noise levels and by analyzing real expression data. For both the theoretical analysis and experimental data, we observe that, under reasonable noise levels, autocorrelations in the time series data allow dense sampling to better determine the correct levels of non-sampled points when compared to replicate sampling. A Java implementation of our framework can be used to determine the best replicate strategy given the expected noise. These results provide theoretical support to the large number of high-throughput time series experiments that do not use replicates.

Deconvolution of Ensemble Chromatin Interaction Data Reveals the Latent Mixing Structures in Cell Subpopulations.

Chromosome conformation capture (3C) experiments provide a window into the spatial packing of a genome in three dimensions within the cell. This structure has been shown to be correlated with gene regulation, cancer mutations, and other genomic functions. However, 3C provides mixed measurements on a population of typically millions of cells, each with a different genome structure due to the fluidity of the genome and differing cell states. Here, we present several algorithms to deconvolve these measured 3C matrices into estimations of the contact matrices for each subpopulation of cells and relative densities of each subpopulation. We formulate the problem as that of choosing matrices and densities that minimize the Frobenius distance between the observed 3C matrix and the weighted sum of the estimated subpopulation matrices. Results on HeLa 5C and mouse and bacteria Hi-C data demonstrate the methods' effectiveness. We also show that domain boundaries from deconvolved matrices are often more enriched or depleted for regulatory chromatin markers when compared to boundaries from convolved matrices.

Resolving spatial inconsistencies in chromosome conformation measurements.

BACKGROUND: Chromosome structure is closely related to its function and Chromosome Conformation Capture (3C) is a widely used technique for exploring spatial properties of chromosomes. 3C interaction frequencies are usually associated with spatial distances. However, the raw data from 3C experiments is an aggregation of interactions from many cells, and the spatial distances of any given interaction are uncertain. RESULTS: We introduce a new method for filtering 3C interactions that selects subsets of interactions that obey metric constraints of various strictness. We demonstrate that, although the problem is computationally hard, near-optimal results are often attainable in practice using well-designed heuristics and approximation algorithms. Further, we show that, compared with a standard technique, this metric filtering approach leads to (a) subgraphs with higher statistical significance, (b) lower embedding error, (c) lower sensitivity to initial conditions of the embedding algorithm, and (d) structures with better agreement with light microscopy measurements. Our filtering scheme is applicable for a strict frequency-to-distance mapping and a more relaxed mapping from frequency to a range of distances. CONCLUSIONS: Our filtering method for 3C data considers both metric consistency and statistical confidence simultaneously resulting in lower-error embeddings that are biologically more plausible.

Parsimonious reconstruction of network evolution.

BACKGROUND: Understanding the evolution of biological networks can provide insight into how their modular structure arises and how they are affected by environmental changes. One approach to studying the evolution of these networks is to reconstruct plausible common ancestors of present-day networks, allowing us to analyze how the topological properties change over time and to posit mechanisms that drive the networks' evolution. Further, putative ancestral networks can be used to help solve other difficult problems in computational biology, such as network alignment. RESULTS: We introduce a combinatorial framework for encoding network histories, and we give a fast procedure that, given a set of gene duplication histories, in practice finds network histories with close to the minimum number of interaction gain or loss events to explain the observed present-day networks. In contrast to previous studies, our method does not require knowing the relative ordering of unrelated duplication events. Results on simulated histories and real biological networks both suggest that common ancestral networks can be accurately reconstructed using this parsimony approach. A software package implementing our method is available under the Apache 2.0 license at http://cbcb.umd.edu/kingsford-group/parana. CONCLUSIONS: Our parsimony-based approach to ancestral network reconstruction is both efficient and accurate. We show that considering a larger set of potential ancestral interactions by not assuming a relative ordering of unrelated duplication events can lead to improved ancestral network inference.