Research in Computational Molecular Biology: 27th Annual International Conference, RECOMB 2023, Istanbul, Turkey, April 16–19, 2023, Proceedings (L Book 13976)

This book constitutes the refereed proceedings of the 27th Annual International Conference on Research in Computational Molecular Biology, RECOMB 2023, held in Istanbul, Turkey, during April 16–19, 2023. The 11 regular and 33 short papers presented in this book were carefully reviewed and selected from 188 submissions. The papers report on original research in all areas of computational molecular biology and bioinformatics. Preface Organization Contents Extended Abstracts VStrains: De Novo Reconstruction of Viral Strains via Iterative Path Extraction from Assembly Graphs 1 Introduction 2 Methods 2.1 Preliminary 2.2 Algorithm Overview 2.3 Preprocessing 2.4 Graph Disentanglement 2.5 Contig-Based Path Extraction 3 Experimental Setup 3.1 Experimental Datasets 3.2 Baselines and Evaluation Metrics 4 Experimental Results 4.1 Performance on Simulated Datasets 4.2 Performance on Real Datasets 5 Software and Resource Usage 6 Conclusion and Discussion References Spectrum Preserving Tilings Enable Sparse and Modular Reference Indexing 1 Introduction 2 Problem Definition and Preliminaries 3 Spectrum Preserving Tilings 3.1 Definition 3.2 A General and Modular Index over Spectrum Preserving Tilings 3.3 ``Drop in'' Implementations for Efficient k-mer-to-tile Queries 3.4 Challenges of the Tile-to-Occurrence Query 4 Pufferfish2 4.1 Interpreting pufferfish as an Index over a Unitig-Based SPT 4.2 Sampling Unitigs and Traversing Tilings to Sparsify the Unitig-to-Occurrence Query 4.3 Implementing the pred Query with pufferfish2 4.4 A Random Sampling Scheme to Guarantee Short Backwards Traversals 4.5 Closing the Gap Between a Constant Time pred Query and Contiguous Array Access 5 Experiments 6 Discussion and Future Work References Statistically Consistent Rooting of Species Trees Under the Multispecies Coalescent Model 1 Introduction 2 Background 2.1 Allman, Degnan, and Rhodes (ADR) Theory 2.2 Quintet Rooting 3 QR-STAR 3.1 Determining the Rooted Shape 4 Theoretical Results 4.1 Statistical Consistency 5 Experimental Study 6 Conclusion References Sequence to Graph Alignment Using Gap-Sensitive Co-linear Chaining 1 Introduction 2 Concepts and Notations 2.1 Co-linear Chaining on Sequences Revisited 2.2 Sparse Dynamic Programming on DAGs Using Minimum Path Cover 3 Problem Formulations 4 Proposed Algorithms 5 Implementation Details 6 Experiments References DM-Net: A Dual-Model Network for Automated Biomedical Image Diagnosis 1 Introduction 2 Methodology 2.1 The Shallow CNN Model: L-Net 2.2 The Deep CNN Model: R-Net 2.3 Loss Function 3 Experiments 3.1 Datasets and Settings 3.2 Segmentation Results 3.3 Ablative Evaluation on Segmentation 4 Conclusions References MTGL-ADMET: A Novel Multi-task Graph Learning Framework for ADMET Prediction Enhanced by Status-Theory and Maximum Flow 1 Introduction 2 Methodology 2.1 Problem Formulation 2.2 Overview of Framework 2.3 Auxiliary Task Selection 2.4 Multi-task Graph Learning Model 3 Experiments 3.1 Dataset and Setup 3.2 Comparisons with State-of-the-Art 3.3 Ablation Studies 3.4 Case Study: Interpretability of MTGL-ADMET 4 Conclusions References CDGCN: Conditional de novo Drug Generative Model Using Graph Convolution Networks 1 Introduction 2 Materials and Methods 2.1 Molecular Graph Generation 2.2 Model Architecture 2.3 Loss Computation and Training 3 Results and Discussion 3.1 Datasets 3.2 Baselines 3.3 Implementation Details 3.4 Evaluation Metrics 3.5 Experimental Evaluation 4 Conclusion References Percolate: An Exponential Family JIVE Model to Design DNA-Based Predictors of Drug Response 1 Introduction 2 Methods 2.1 Trade-off Between Robust and Predictive Types 2.2 Exponential Family Distribution 2.3 Saturated Model Parameters 2.4 Generalized Linear Model Principal Component Analysis (GLM-PCA) 2.5 Comparison of GLM-PCA Directions by Percolate 2.6 Projector of Joint Signal 2.7 Drug Response Prediction 2.8 Data Download, Modelling and Processing 3 Results 3.1 The Breakdown of the Joint Signals Highlights the Topology of Multi-omics Data 3.2 Robust Signal Predictive of Drug Response Is Concentrated in the Joint Part 3.3 Out-of-sample Extension Recapitulates the Predictive Performance of Robust Signal 3.4 Study of Genes Contributing to the Joint Signals 3.5 Iterative Application of Percolate Deprives Gene Expression from Predictive Power 4 Discussion References Translation Rate Prediction and Regulatory Motif Discovery with Multi-task Learning 1 Introduction 2 Methods 2.1 MTtrans Model 2.2 Extraction of Sequence Motifs from Convolutional Filters 2.3 Motif Similarity Comparison 2.4 Motifs Matching 2.5 Building Logistic Regression and Random Forest Model on PWM-Derived Scores 2.6 Identification of Important Motifs 3 Results 3.1 MTtrans Learns the Shared Patterns from Multiple Experimental Systems 3.2 MTtrans Better Coordinates MPRA Tasks and Improves Translation Rate Prediction 3.3 MTtrans is Robust Across Replicates 3.4 MTtrans Learns More Transferable Sequence Features 3.5 MTtrans Better Predicts the Translation Rate of Endogenous Transcripts in Human Cell Lines 3.6 Discovery of 5'UTR Sequence Motifs from the Deeper Layer of Shared Encoder 3.7 The Discovered Regulatory Motifs can be Experimentally Validated 4 Discussion A Appendix References Computing Shortest Hyperpaths for Pathway Inference in Cellular Reaction Networks 1 Introduction 2 Shortest Hyperpaths in Directed Hypergraphs 3 Computing Hyperpaths by Integer Programming 3.1 Characterizing Superpaths via Cuts 3.2 Representing Superpaths by Linear Inequalities 3.3 Solving the Integer Program by a Cutting Plane Algorithm 3.4 Strengthening the Initial Integer Program 3.5 A Separation Algorithm Leveraging Distance-Based Cuts 4 Experimental Results 4.1 Experimental Setup 4.2 Comparing Alternate Hyperpath Methods 4.3 Speed of Computing Optimal Hyperpaths 4.4 A Concrete Biological Example 4.5 Analysis of Recovering Known Pathways 5 Conclusion References T-Cell Receptor Optimization with Reinforcement Learning and Mutation Polices for Precision Immunotherapy 1 Introduction 2 Related Work 3 Methods 3.1 Problem Definition 3.2 Mutation Policy Network 3.3 Potential TCR Validity Measurement 3.4 TCRPPO Learning 4 Experimental Settings 4.1 Datasets 4.2 Experimental Setup 4.3 Baseline Methods 4.4 Evaluation Metrics 5 Experimental Results 5.1 Comparison on TCR Optimization Methods 5.2 Evaluation on Optimized TCR Sequences 6 Conclusions and Outlook References Short Papers TREE-QMC: Improving Quartet Graph Construction for Scalable and Accurate Species Tree Estimation from Gene Trees References mapquik: Efficient Low-Divergence Mapping of Long Reads in Minimizer Space Deriving Confidence Intervals for Mutation Rates Across a Wide Range of Evolutionary Distances Using FracMinHash References Entropy Predicts Sensitivity of Pseudo-random Seeds Reference Seed-Chain-Extend Alignment is Accurate and Runs in Close to O(m logn) Time for Similar Sequences: A Rigorous Average-Case Analysis References Extremely-Fast Construction and Querying of Compacted and Colored de Bruijn Graphs with GGCAT References PASTE2: Partial Alignment of Multi-slice Spatially Resolved Transcriptomics Data References FastRecomb: Fast Inference of Genetic Recombination Rates in Biobank Scale Data References Efficient Taxa Identification Using a Pangenome Index 1 Introduction References Vector-Clustering Multiple Sequence Alignment: Aligning into the Twilight Zone of Protein Sequence Similarity with Protein Language Models Single-Cell Methylation Sequencing Data Reveal Succinct Metastatic Migration Histories and Tumor Progression Models References Information-Theoretic Classification Accuracy: A Criterion That Guides Data-Driven Combination of Ambiguous Outcome Labels in Multi-class Classification Reference Efficient Minimizer Orders for Large Values of k Using Minimum Decycling Sets References Dashing 2: Genomic Sketching with Multiplicities and Locality-Sensitive Hashing 1 Introduction 2 Methods Summary 3 Results Summary References Startle: A Star Homoplasy Approach for CRISPR-Cas9 Lineage Tracing A Fast and Scalable Method for Inferring Phylogenetic Networks from Trees by Aligning Lineage Taxon Strings (Extended Abstract) References Aligning Distant Sequences to Graphs Using Long Seed Sketches References MD-Cat: Phylogenetic Dating Under a Flexible Categorical Model Using Expectation-Maximization Phenotypic Subtyping via Contrastive Learning HOGVAX: Exploiting Peptide Overlaps to Maximize Population Coverage in Vaccine Design with Application to SARS-CoV-2 Ultra-Fast Genome-Wide Inference of Pairwise Coalescence Times References Leveraging Family Data to Design Mendelian Randomization That is Provably Robust to Population Stratification References Minimal Positional Substring Cover: A Haplotype Threading Alternative to Li & Stephens Model References Cell Segmentation for High-Resolution Spatial Transcriptomics References Unsupervised Deep Peak Caller for ATAC-seq References Unraveling Causal Gene Regulation from the RNA Velocity Graph Using Velorama PIsToN: Evaluating Protein Binding Interfaces with Transformer Networks References DebiasedDTA: A Framework for Improving the Generalizability of Drug-Target Affinity Prediction Models 1 Introduction 2 DebiasedDTA 3 Results 4 Conclusion References Drug Synergistic Combinations Predictions via Large-Scale Pre-training and Graph Structure Learning 1 Introduction 2 Methods 3 Results 4 Conclusion References Pisces: A Cross-Modal Contrastive Learning Approach to Synergistic Drug Combination Prediction Modeling and Predicting Cancer Clonal Evolution with Reinforcement Learning References Enabling Trade-Offs in Privacy and Utility in Genomic Data Beacons and Summary Statistics References Accurate Evaluation of Transcriptomic Re-identification Risks Using Discriminative Sequence Models 1 Introduction 2 Methods 3 Results References Author Index