Investigating the Role of Pangolin Pseudogenes in Host-Pathogen Immune Interactions: A BiLSTM-Based Approach

Abstract

This study integrates a Bidirectional Long Short-Term Memory (BiLSTM) deep learning framework with bioinformatics approaches to elucidate the functional role of pangolin pseudogenes in the immune interactions between tick-borne pathogens and their hosts. We developed a novel BiLSTM-autoencoder model incorporating dynamic weight positional encoding and a multi-head self-attention mechanism, effectively capturing pseudogene sequence features while overcoming limitations inherent in traditional analytical methods. Genome-wide screening of the Manis pentadactyla (GCF_030020395.1) and Manis javanica (GCF_001685135.1) assemblies identified 3,209 and 2,035 pseudogenes, respectively. Subsequent filtration yielded 94 immune-related homologous genes, classified into ten distinct immune system pathways. Our analysis reveals significant species-specific variation and functional plasticity in pangolin pseudogenes: Chinese Pangolin LILRA6 Pseudogenes: Fourteen identified variants modulate bacterial recognition and inflammatory responses through specific amino acid deletions. Interferon Receptor Pseudogenes (e.g., XP_036870501.1): Critical mutations disrupt JAK-STAT signaling pathway functionality. Toll-like Receptor (TLR) Pseudogenes: Reconstruction error values (ranging from 28.149 to 68.957) correlate strongly with structural integrity. Notably, the disrupted LRR domain in the giant pangolin sequence KAK2502118.1 (error = 65.986) suggests an adaptive immune strategy. Furthermore, we identified a potential molecular interaction between a pangolin-derived 52 kDa sequence and the 8.9 kDa salivary protease of Amblyomma javanense, providing novel insights into tick-borne transmission mechanisms. This study represents the first application of deep learning to elucidate the functional role of pangolin pseudogenes, confirming their active involvement in immune resistance. It establishes a new paradigm for investigating host-pathogen interactions and provides a critical foundation for the analysis of underlying data in the surveillance and control of tick-borne diseases.