MuSL: Multimodal deep learning for generalizable prediction of synthetic lethality from sequence, transcriptomic, and network data Article Swipe

MAKE IMPACT

Fang Yimiao , Wang Jie , Hong, Mutian , Zheng Jie · YOU? · · 2025 · Open Access · · DOI: https://doi.org/10.5281/zenodo.17098066

MuSL: Multimodal deep learning for generalizable prediction of synthetic lethality from sequence, transcriptomic, and network data Model Architecture The MuSL framework integrates multimodal biological data through a tri-branch deep learning architecture to predict synthetic lethal (SL) gene pairs. As illustrated in the figure above, the model jointly learns from transcriptomic images, statistical features, and protein interaction networks. 1. Feature Learning Pathways MuSL processes input data through three complementary pathways: Image-Based Expression Branch (CNN): Transforms the transcriptome profiles of gene pairs into two-dimensional joint density distribution maps (32×3232×32 histograms). Utilizes a Convolutional Neural Network (CNN) to automatically extract spatial features, capturing complex patterns such as functional decoupling and mutual exclusivity directly from raw data. Statistical Feature Branch (MLP): Extracts 35 handcrafted statistical features capturing explicit expression patterns (e.g., correlation, variation) from the same transcriptomic data. Projects these features through a dedicated fully connected layer (MLP) to form a representation that complements the deep image features. Graph-Based Network Branch (GNN): Models the Protein-Protein Interaction (PPI) network using a Graph Neural Network (GNN). Crucial Innovation: Initializes node features using ESM2 protein language model embeddings, integrating evolutionary and sequence-level priors to generate robust topological relationship features. 2. Adaptive Fusion & Prediction Cross-Modal Integration: The framework employs a cross-attention and gating fusion mechanism to dynamically weight and integrate features from all three pathways (CNN, Statistics, and GNN) based on their contextual importance. Multi-Head Prediction: Predictions are generated from both modality-specific heads and the final fused representation using four specialized classifiers. 3. Training Strategy Composite Loss Function: The model optimizes a joint objective that combines: Classification losses from all four prediction heads. Contrastive learning components to enforce semantic alignment and improve cross-modal consistency. Required Data Files The following H5AD files need to be downloaded and placed in the processed_data/ directory: a549_cell_line_imputed.h5ad - A549 cell line single-cell RNA-seq data k562_cell_line_imputed.h5ad - K562 cell line single-cell RNA-seq data tcga_all.h5ad - TCGA bulk RNA-seq data protein_embeddings.pt - Embeddings from ESM2 all_emb_genept.pkl - Embeddings from GenePT

Type

dataset

Landing Page

https://doi.org/10.5281/zenodo.17098066

OA Status

green

OpenAlex ID

https://openalex.org/W7111412325

Raw OpenAlex JSON

OpenAlex ID

https://openalex.org/W7111412325

Canonical identifier for this work in OpenAlex

DOI

https://doi.org/10.5281/zenodo.17098066

Digital Object Identifier

Title

MuSL: Multimodal deep learning for generalizable prediction of synthetic lethality from sequence, transcriptomic, and network data

Work title

Type

dataset

OpenAlex work type

Publication year

2025

Year of publication

Publication date

2025-12-02

Full publication date if available

Authors

Fang Yimiao, Wang Jie, Hong, Mutian, Zheng Jie

List of authors in order

Landing page

https://doi.org/10.5281/zenodo.17098066

Publisher landing page

Open access

Yes

Whether a free full text is available

OA status

green

Open access status per OpenAlex

OA URL

https://doi.org/10.5281/zenodo.17098066

Direct OA link when available

Concepts

Computer science, Artificial intelligence, Deep learning, Convolutional neural network, Feature learning, Machine learning, Fusion mechanism, Feature (linguistics), Representation (politics), Network architecture, Synthetic lethality, Feature engineering, Statistical model, Graph, Multimodal learning, Pattern recognition (psychology), ENCODE, Data modeling, Artificial neural network, Synthetic data, Network model, Deep belief network, Data mining, Joint probability distribution, Node (physics), Interaction network, Robustness (evolution), Gene regulatory network, External Data Representation, Feature extraction, Prior probability, Statistical relational learning, Feature vector, Raw data, Subspace topology, Visualization

Top concepts (fields/topics) attached by OpenAlex

Cited by

0

Total citation count in OpenAlex

Full payload