<i>Deep learning (DL) methods show promising potential for single-cell data analysis, yet required tremendous efforts in building the models. </i><i>To streamline the application of sequence-based DL methods in single-cell genomics, we established a two-layer CNN model as a baseline model and systematically evaluate how data characteristics, hyperparameter optimization, and advanced model architectures affect performance in sequence-to-expression and sequence-to-regulation tasks. We further explored the application of multi-task learning (MTL) frameworks for modeling cellular heterogeneity, evaluating the effectiveness of task grouping and balancing strategies, with particular focus on the prediction of rare cell types. </i><i>Our comprehensive benchmark efforts provide an actionable framework and valuable insights for guiding future research endeavors and facilitating the development of the sequence-based DL models capable of superior predictive performance in single-cell genomics.</i>
Precise and interpretable classification of autism-related behaviors is importantfor initial diagnosis, personalized intervention, and support arrangements. This studyproposes an interpretable machine learning (ML) model using Light GradientBoosting Machine (LightGBM) and Categor…
<b>Abstract:</b>Modern real estate platforms manage heterogeneous buyer populations ranging from first-time residential home buyers to institutional corporate entities and international high-net-worth investors. Traditional marketing strategies relying on broad demographic genera…
Background & Objective: Chronic sleep deprivation among adolescents has accelerated over the past three decades, aligning with widespread digital media saturation and reported cognitive focus issues. Traditional public health tracking often evaluates short‑term trends, overlo…
Leaf Area Index (LAI) serves as a key biophysical parameter for characterizing vegetation canopy structure and ecosystem functions. To address the absence of LAI products for the Fengyun-3B (FY-3B) satellite and the limitations of current satellite LAI products, this study propos…
Record of the development of the deep learning pipeline CADENCE, capable of generating novel competitive kinase inhibitors from amino acid sequence. CADENCE was trained on the Davis dataset of kinase inhibitors and binding scores, creating a binding affinity prediction model. Thi…
"MachineLearning_mono" is a MATLAB code that can read the FCC and BCC dataset, train the Machine Learning models, and generate plots for the performance of the ML models and make predictions. "BayeSQP_optim" is a MATLAB code that can take the initial guess from ML and do optimiza…