Biology subjects

Xu, D.

Publications and source records attributed to Xu, D..

At least 109 records · Page 6Linked to original sources

bioRxiv2024

Medium Composition Affects Microbial Corrosion Rates

Desulfovibrio vulgaris and Desulfovibrio ferrophilus were previously proposed to have distinct iron corrosion mechanisms because D. ferrophilus corroded faster. However, the chloride concentration in the D. ferrophilus marine medium was much higher than in the D. vulgaris freshwater medium. D. vulgaris corrosion rates accelerated with increasing chloride and were faster than D. ferrophilus in the same marine medium. Differences in D. ferrophilus corrosion rates in two different media with the same chloride concentration suggested that minor differences in other medium constituents also impact on microbial corrosion. These results demonstrate the importance of considering medium composition in microbial corrosion studies.

microbiology↗

bioRxiv2024

Elucidating Microbial Iron Corrosion Mechanisms with a Hydrogenase-Deficient Strain of Desulfovibrio vulgaris

Sulfate-reducing microorganisms extensively contribute to the corrosion of ferrous metal infrastructure. There is substantial debate over their corrosion mechanisms. We investigated Fe0 corrosion with Desulfovibrio vulgaris, the sulfate reducer most often employed in corrosion studies. Cultures were grown with both lactate and Fe0 as potential electron donors to replicate the common environmental condition in which organic substrates help fuel the growth of corrosive microbes. Fe0 was corroded in cultures of a D. vulgaris hydrogenase-deficient mutant with the 1:1 correspondence between Fe0 loss and H2 accumulation expected for Fe0 oxidation coupled to H+ reduction to H2. This result and the extent of sulfate reduction indicated that D. vulgaris was not capable of direct Fe0-to-microbe electron transfer even though it was provided with a supplementary energy source in the presence of abundant ferrous sulfide. Corrosion in the hydrogenase-deficient mutant cultures was greater than in sterile controls, demonstrating the H2 removal was not necessary for the enhanced corrosion observed in the presence of microbes. The parental H2-consuming strain corroded more Fe0 than the mutant strain, which could be attributed to H2 oxidation coupled to sulfate reduction producing sulfide that further stimulated Fe0 oxidation. The results suggest that H2 consumption is not necessary for microbially enhanced corrosion, but H2 oxidation can indirectly promote corrosion by increasing sulfide generation from sulfate reduction. The finding that, D. vulgaris was incapable of direct electron uptake from Fe0 reaffirms that direct metal-to-microbe electron transfer has yet to be rigorously described in sulfate-reducing microbes. Impact StatementThe economic impact of microbial corrosion of iron-containing metals is substantial. A better understanding of how microbes accelerate corrosion is expected to lead to the development of methods to prevent corrosion. The results presented here refute the suggestion, frequently made in the microbiology literature, that microbial H2 uptake stimulates Fe0 corrosion. Also refuted, are previous claims that Desulfovibrio vulgaris is capable of directly extracting electrons from Fe0. The results are consistent with the concept that sulfide produced by sulfate-reducers promotes Fe0 oxidation with the production of H2. The results illustrate that appropriate mutants can provide rigor to corrosion mechanism studies.

microbiology↗

bioRxiv2024

A PTER-dependent pathway of taurine metabolism linked to energy balance

Taurine is a conditionally essential micronutrient and one of the most abundant amino acids in humans1-3. In endogenous taurine metabolism, dedicated enzymes are involved in biosynthesis of taurine from cysteine as well as the downstream derivatization of taurine into secondary taurine metabolites4,5. One such taurine metabolite is N-acetyltaurine6. Levels of N-acetyltaurine are dynamically regulated by diverse physiologic perturbations that alter taurine and/or acetate flux, including endurance exercise7, nutritional taurine supplementation8, and alcohol consumption6,9. While taurine N-acetyltransferase activity has been previously detected in mammalian cells6,7, the molecular identity of this enzyme, and the physiologic relevance of N-acetyltaurine, have remained unknown. Here we show that the orphan body mass index-associated enzyme PTER (phosphotriesterase-related)10 is the principal mammalian taurine N-acetyltransferase/hydrolase. In vitro, recombinant PTER catalyzes bidirectional taurine N-acetylation with free acetate as well as the reverse N-acetyltaurine hydrolysis reaction. Genetic ablation of PTER in mice results in complete loss of tissue taurine N-acetyltransferase/hydrolysis activities and systemic elevation of N-acetyltaurine levels. Upon stimuli that increase taurine levels, PTER-KO mice exhibit lower body weight, reduced adiposity, and improved glucose homeostasis. These phenotypes are recapitulated by administration of N-acetyltaurine to wild-type mice. Lastly, the anorexigenic and anti-obesity effects of N-acetyltaurine require functional GFRAL receptors. Together, these data uncover enzymatic control of a previously enigmatic pathway of secondary taurine metabolism linked to energy balance.

biochemistry↗

bioRxiv2024

Profound synthetic lethality between SMARCAL1 and FANCM

DNA replication stress is a threat to genome integrity. The large SNF2-family of ATPases participates in preventing and mitigating DNA replication stress by employing their ATP-driven motor to remodel DNA or DNA-bound proteins. To understand the contribution of these ATPases in genome maintenance, we undertook CRISPR-based synthetic lethality screens with three SNF2-type ATPases: SMARCAL1, ZRANB3 and HLTF. Here we show that SMARCAL1 displays a profound synthetic lethal interaction with FANCM, another ATP-dependent translocase involved in DNA replication and genome stability. Their combined loss causes severe genome instability that we link to chromosome breakage at loci enriched in simple repeats, which are known to challenge replication fork progression. Our findings illuminate a critical genetic buffering mechanism that provides an essential function for maintaining genome integrity.

molecular biology↗

bioRxiv2024

Arabidopsis GENOMES UNCOUPLED PROTEIN1 binds to plastid RNAs and promotes their maturation

Plastid biogenesis and the coordination of plastid and nuclear genome expression through anterograde and retrograde signaling are essential for plant development. GENOMES UNCOUPLED1 (GUN1) plays a central role in retrograde signaling during early plant development. The putative function of GUN1 has been extensively studied, but its molecular function remains controversial. Here, we evaluate published transcriptome data and generate our own data from gun1 mutants grown under signaling relevant conditions to show that editing and splicing are not relevant for GUN1-dependent retrograde signaling. Our study of the plastid (post)-transcriptome of gun1 seedlings with white and pale cotyledons demonstrates that GUN1 deficiency significantly alters the entire plastid transcriptome. By combining this result with a PPR code-based prediction and experimental validation by RNA immunoprecipitation experiments, several targets of GUN1 were identified, including 23S rRNA, tRNAs and RNAs derived from ycf1.2 and the ndhH-ndhA-ndhI-ndhG-ndhE-psaC-ndhD gene cluster. The absence of plastid rRNAs and the significant reduction of almost all plastid transcripts in white gun1 mutants account for the cotyledon phenotype. Our study identifies RNA binding and maturation as the long-sought molecular function of GUN1 and resolves long-standing controversies. We anticipate that our findings will serve as a basis for subsequent studies investigating the mechanism of plastid gene expression and will facilitate the elucidation of GUN1s function in retrograde signaling.

plant biology↗

bioRxiv2024

Parameter-Efficient Fine-Tuning Enhances Adaptation of Single Cell Large Language Model for Cell Type Identification

Single-cell sequencing transformed biology and medicine, providing an unprecedented high-resolution view at the cellular level. However, the vast variability inherent in single-cell sequencing data impedes its utility for in-depth downstream analysis. Inspired by the foundation models in natural language processing, recent advancements have led to the development of single-cell Large Language Models (scLLMs). These models are designed to discern universal patterns across diverse single-cell datasets, thereby enhancing the signal-to-noise ratio. Despite their potential, multiple studies indicate existing scLLMs do not perform well in zero-short settings, highlighting a pressing need for more effective adaptation techniques. This research proposes several adaptation techniques for scLLMs by preserving the original model parameters while selectively updating newly introduced tensors. This approach aims to overcome the limitations associated with traditional fine-tuning practices, such as catastrophic forgetting and computational inefficiencies. We introduce two Parameter-Efficient Fine-Tuning (PEFT) strategies specifically tailored to refine scLLMs for cell type identification. Our investigations utilizing scGPT demonstrate that PEFT can enhance performance, with the added benefit of up to a 90% reduction in parameter training compared to conventional fine-tuning methodologies. This work paves the way for a new direction in leveraging single-cell models with greater efficiency and efficacy in single-cell biology.

bioinformatics↗

bioRxiv2024

A Comprehensive Evaluation of Large Language Models in Mining Gene Interactions and Pathway Knowledge

BackgroundUnderstanding complex biological pathways, including gene-gene interactions and gene regulatory networks, is critical for exploring disease mechanisms and drug development. Manual literature curation of biological pathways is useful but cannot keep up with the exponential growth of the literature. Large-scale language models (LLMs), notable for their vast parameter sizes and comprehensive training on extensive text corpora, have great potential in automated text mining of biological pathways. MethodThis study assesses the effectiveness of 21 LLMs, including both API-based models and open-source models. The evaluation focused on two key aspects: gene regulatory relations (specifically, activation, inhibition, and phosphorylation) and KEGG pathway component recognition. The performance of these models was analyzed using statistical metrics such as precision, recall, F1 scores, and the Jaccard similarity index. ResultsOur results indicated a significant disparity in model performance. Among the API-based models, ChatGPT-4 and Claude-Pro showed superior performance, with an F1 score of 0.4448 and 0.4386 for the gene regulatory relation prediction, and a Jaccard similarity index of 0.2778 and 0.2657 for the KEGG pathway prediction, respectively. Open-source models lagged their API-based counterparts, where Falcon-180b-chat and llama1-7b led with the highest performance in gene regulatory relations (F1 of 0.2787 and 0.1923, respectively) and KEGG pathway recognition (Jaccard similarity index of 0.2237 and 0. 2207, respectively). ConclusionLLMs are valuable in biomedical research, especially in gene network analysis and pathway mapping. However, their effectiveness varies, necessitating careful model selection. This work also provided a case study and insight into using LLMs as knowledge graphs.

bioinformatics↗

bioRxiv2024

Deep Learning Analysis on Images of iPSC-derived Motor Neurons Carrying fALS-genetics Reveals Disease-Relevant Phenotypes

Amyotrophic lateral sclerosis (ALS) is a devastating condition with very limited treatment options. It is a heterogeneous disease with complex genetics and unclear etiology, making the discovery of disease-modifying interventions very challenging. To discover novel mechanisms underlying ALS, we leverage a unique platform that combines isogenic, induced pluripotent stem cell (iPSC)-derived models of disease-causing mutations with rich phenotyping via high-content imaging and deep learning models. We introduced eight mutations that cause familial ALS (fALS) into multiple donor iPSC lines, and differentiated them into motor neurons to create multiple isogenic pairs of healthy (wild-type) and sick (mutant) motor neurons. We collected extensive high-content imaging data and used machine learning (ML) to process the images, segment the cells, and learn phenotypes. Self-supervised ML was used to create a concise embedding that captured significant, ALS-relevant biological information in these images. We demonstrate that ML models trained on core cell morphology alone can accurately predict TDP-43 mislocalization, a known phenotypic feature related to ALS. In addition, we were able to impute RNA expression from these image embeddings, in a way that elucidates molecular differences between mutants and wild-type cells. Finally, predictors leveraging these embeddings are able to distinguish between mutant and wild-type both within and across donors, defining cellular, ML-derived disease models for diverse fALS mutations. These disease models are the foundation for a novel screening approach to discover disease-modifying targets for familial ALS.

neuroscience↗

bioRxiv2024

Heparan sulfate selectively inhibits the collagenase activity of cathepsin K

Cathepsin K (CtsK) is a cysteine protease with potent collagenase activity. CtsK is highly expressed by bone-resorbing osteoclasts and plays an essential role in bone remodeling. Although CtsK is known to bind heparan sulfate (HS), the structural details of the interaction, and how HS ultimately regulates the biological functions of CtsK, remains largely unknown. In this report, we determined that CtsK preferably binds to larger HS oligosaccharides, such as dodecasaccharides (12mer), and that the12mer can induce monomeric CtsK to form a stable dimer in solution. Interestingly, while HS has no effect on the peptidase activity of CtsK, it greatly inhibits the collagenase activity of CtsK in a manner dependent on sulfation level. By forming a complex with CtsK, HS was able to preserve the full peptidase activity of CtsK for prolonged periods, likely by stabilizing its active conformation. Crystal structures of Ctsk with a bound 12mer, alone and in the presence of the endogenous inhibitor cystatin-C reveal the location of HS binding is remote from the active site. Mutagenesis based on these complex structures identified 6 basic residues of Ctsk that play essential roles in mediating HS-binding. At last, we show that HS 12mers can effectively block osteoclast resorption of bone in vitro. Combined, we have shown that HS can function as a multifaceted regulator of CtsK and that HS-based oligosaccharide might be explored as a new class of selective CtsK inhibitor in many diseases that involve exaggerated bone resorption.

biochemistry↗

bioRxiv2023

CentIER: accurate centromere identification for plant genomes with sequence specificity information

Centromere identification is one of the important problems in genomics, providing a foundation for the studies of centromeres in aspects of composition, functionality, evolution, inheritance, and variation. The existing wet-experiment-based method is costly and time-consuming, while the bioinformatic method can only detect tandem repeats losing non-repetitive sequence regions in the centromere. To address these shortcomings, we introduce a new pipeline, CentIER, for the automatic and accurate identification and annotation of centromere regions by taking advantage of the sequence specificity information. CentIER only requires users to input the genomic sequence, and then it can partition the centromeric region from a chromosome, identify tandem repeat monomers, annotate retrotransposons, and ultimately output visualized results. By referencing the experimentally determined centromere regions, it was discovered that the predictive accuracy of centromere recognition by CentIER exceeded 90%. Following the evaluation of CentIERs accuracy, it was applied to investigate the sequence and distribution characteristics of centromeric retrotransposons and tandem repeat sequences of different species, providing insights into these traits in monocotyledonous and dicotyledonous plants.

bioinformatics↗

bioRxiv2023

Iterative Prompt Refinement for Mining Gene Relationships from ChatGPT

ChatGPT has demonstrated its potential as a surrogate knowledge graph. Trained on extensive data sources, including open-access publications, peer-reviewed research articles and biomedical websites, ChatGPT extracted information on gene relationships and biological pathways. However, a major challenge is model hallucination, i.e., high false positive rates. To assess and address this challenge, we systematically evaluated ChatGPTs capacity for predicting gene relationships using GPT-3.5-turbo and GPT-4. Benchmarking against the KEGG Pathway Database as the ground truth, we experimented with diverse prompting strategies, targeting gene relationships of activation, inhibition, and phosphorylation. We introduced an innovative iterative prompt refinement technique. By assessing prompt efficacy using metrics like F-1 score, precision, and recall, GPT-4 was re-engaged to suggest improved prompts. A refined prompt, which combines a specialized role with explanatory text, significantly enhances the performance. Going beyond pairwise gene relationships, we also deciphered complex gene interplays, such as gene interaction chains and pathways pertinent to diseases like non-small cell lung cancer. Direct prompts showed limited success, but "least-to-most" prompting exhibited significant potentials for such network constructions. The methods in this study may be used for some other bioinformatics prediction problems.

bioinformatics↗

bioRxiv2023

Inhibition of KLK8 promotes pulmonary endothelial repair by restoring VE-cadherin/Akt/FOXM1 pathway

The tissue kallikrein-related peptidases (KLKs) are secreted serine proteases deeply involved in angiogenesis. However, whether KLKs are involved in the regulation of endothelial regeneration during sepsis remains unknown. By comparing the mRNA levels of 15 KLKs, we found that KLK8 was the highest induced KLK member in lung tissues or primary isolated mouse lung vascular endothelial cells (MLVECs) exposed to lipopolysaccharide (LPS). Adenovirus-mediated overexpression of KLK8 caused endothelial hyperpermeability both in vitro and in vivo. Inhibition of KLK8, by either gene knockout or KLK8 neutralizing antibodies, alleviated sepsis-induced endothelial hyperpermeability, acute lung injury and mortality. Mechanistically, transcription profiling of KLK8-overexpressed endothelial cells revealed a central role of forkhead box M1 (FOXM1) downregulation in mediating the pro-injury and anti-proliferation effects of KLK8. KLK8 cleaved VE-cadherin and consequently suppressed FOXM1 expression by inactivation of the VE-cadherin/Akt pathway. KLK8 deficiency or blockade rescued VE-cadherin/Akt/FOXM1 pathway, thus promoting endothelium regeneration. This study reveals a critical role for KLK8-induced inactivation of VE-cadherin/Akt/FOXM1 pathway in mediating the impairment of endothelial regeneration and the consequent lung vascular leakiness in response to sepsis. HighlightsO_LIUpregulated KLK8 mediates lung endothelial barrier dysfunction during sepsis C_LIO_LIKLK8 inactivates VE-cadherin/Akt/FOXM1, thus impairing endothelium regeneration C_LIO_LIKLK8 deficiency or blockade rescues VE-cadherin/Akt/FOXM1 signaling pathway C_LIO_LIKLK8 deficiency or blockade promotes endothelium regeneration during sepsis C_LIO_LIKLK8 deficiency or blockade attenuates sepsis-induced acute lung injury and mortality C_LI

physiology↗

bioRxiv2023

Structural variation discovery in wheat using PacBio high-fidelity sequencing

BackgroundStructural variations (SVs) pervade plant genomes and contribute substantially to the phenotypic diversity. However, most SVs were ineffectively assayed because of their complex nature and the limitations of early genomic technologies. The recent advance in third-generation sequencing, particularly the PacBio high-fidelity (HiFi) sequencing technology, produces highly accurate long-reads and offers an unprecedented opportunity to characterize SVs structure and functionality. As HiFi sequencing is relatively new to population genomics, it is imperative to evaluate and optimize HiFi sequencing based SV detection before applying the technology at scale. ResultsWe sequenced wheat genomes using HiFi reads, followed by a comprehensive evaluation of mainstream long-read aligners and SV callers in SV detection. The results showed that the accuracy of deletion discovery is markedly influenced by callers, which account for 87.73% of the variance, while both aligners (38.25%) and callers (49.32%) contributed substantially to the accuracy variance for insertions. Among the aligners, Winnowmap2 and NGMLR excelled in detecting deletions and insertions, respectively. For SV callers, SVIM achieved the best performance. We demonstrated that combining the aligners and callers mentioned above is optimal for SV detection. Furthermore, we evaluated the effect of sequencing depth on the accuracy of SV detection, showing that low-coverage HiFi sequencing is sufficiently robust for high-quality SV discovery. ConclusionsThis study thoroughly evaluated SV discovery approaches using HiFi reads, establishing optimal workflows to investigate structural variations in the wheat genome. The notable accuracy of SV discovery from low-coverage HiFi sequencing indicates that skim HiFi sequencing is effective and preferable to characterize SVs at the population level. This study will help advance SV discovery and decipher the biological functions of SVs in wheat and many other plants.

genomics↗

bioRxiv2023

Miffi: Improved accuracy of CNN-based cryo-EM micrograph filtering with fine-tuning and Fourier space information

Efficient and high-accuracy filtering of cryo-electron microscopy (cryo-EM) micrographs is an emerging challenge with the growing speed of data collection and sizes of datasets. Convolutional neural networks (CNNs) are machine learning models that have been proven successful in many computer vision tasks, and have been previously applied to cryo-EM micrograph filtering. In this work, we demonstrate that two strategies, fine-tuning models from pretrained weights and including the power spectrum of micrographs as input, can greatly improve the attainable prediction accuracy of CNN models. The resulting software package, Miffi, is open-source and freely available for public use (https://github.com/ando-lab/miffi).

biophysics↗

bioRxiv2023

Identification of a highly drought-resistant pp7l hda6 mutant

Plants have evolved efficient strategies to cope with drought stress, including stomata closure, significant changes in nuclear gene expression, and epigenetic mechanisms. Previously, we identified Arabidopsis thaliana PROTEIN PHOSPHATASE7-LIKE (PP7L) as an extrachloroplastic protein that promotes chloroplast development and high light, and salt tolerance. Here, we demonstrate that the pp7l mutant can withstand prolonged periods of drought stress. Interestingly, chloroplast development in pp7l recovers under drought conditions, despite growth of the mutant is impaired under normal growth conditions. To assess the (post)transcriptional changes occurring in the pp7l mutant under different durations of drought exposure, we used long non-coding RNA-sequencing. Compared to the previously reported drought-responsive changes in the wild type, the drought-responsive changes detected in the pp7l mutant were negligible. Our analysis of data generated in this study and previously motivated us to create a pp7l hda6 mutant, which exhibits remarkable drought resistance. Notably, the growth penalty associated with pp7l was alleviated in the double mutant, ruling out a dwarf effect on the drought-tolerant trait of this genotype.

plant biology↗

bioRxiv2023

Exploring the Mosaic-like Tissue Architecture of Kidney Diseases Using Relation Equivariant Graph Neural Networks on Spatially Resolved Transcriptomics

Emerging spatially resolved transcriptomics (SRT) technologies provide unprecedented opportunities to discover the spatial patterns of gene expression at the cellular or tissue levels. Currently, most existing computational tools on SRT are designed and tested on the ribbon-like brain cortex. Their present expressive power often makes it challenging to identify highly heterogeneous mosaic-like tissue architectures, such as tissues from kidney diseases. This demands heightened precision in discerning the cellular and morphological changes within renal tubules and their interstitial niches. We present an empowered graph deep learning framework, REGNN (Relation Equivariant Graph Neural Networks), for SRT data analyses on heterogeneous tissue structures. To increase expressive power in the SRT lattice using graph modeling, the proposed REGNN integrates equivariance to handle the rotational and translational symmetries of the spatial space, and Positional Encoding (PE) to identify and strengthen the relative spatial relations of the nodes uniformly distributed in the lattice. Our study finds that REGNN outperforms existing computational tools in identifying inherent mosaic-like heterogenous tissue architectures in kidney samples sourced from different kidney diseases using the 10X Visium platform. In case studies on acute kidney injury and chronic kidney diseases, the results identified by REGNN are also validated by experienced nephrology physicians. This proposed framework explores the expression patterns of highly heterogeneous tissues with an enhanced graph deep learning model, and paves the way to pinpoint underlying pathological mechanisms that contribute to the progression of complex diseases. REGNN is publicly available at https://github.com/Mraina99/REGNN.

bioinformatics↗

bioRxiv2023

Exploring building blocks of cell organization by estimating network motifs using graph isomorphism network

The spatial arrangement of cells within tissues plays a pivotal role in shaping tissue functions. A critical spatial pattern is network motif as the building blocks of cell organization. Network motifs can be represented as recurring significant interconnections of cells with various types in a spatial cell-relation graph, i.e., enriched occurrences of isomorphic subgraphs in the graph, which is computationally infeasible to have an optimal solution with large-size (>3 nodes) subgraphs. We introduce Triangulation Network Motif Neural Network (TrimNN), a neural network-based approach designed to estimate the prevalence of network motifs of any size in a triangulated cell graph. TrimNN simplifies the intricate task of occurrence regression by decomposing it into several binary present/absent predictions on small graphs. TrimNN is trained using representative pairs of predefined subgraphs and triangulated cell graphs to estimate overrepresented network motifs. On typical spatial omics samples within thousands of cells in dozens of cell types, TrimNN robustly infers the presence of a large-size network motif in seconds. In a case study using STARmap Plus technologies, TrimNN identified several biological meaningful large-size network motifs significantly enriched in a mouse model of Alzheimers disease at different months of age. TrimNN provides an accurate, efficient, and robust approach for quantifying network motifs, which helps pave the way to disclose the biological mechanisms underlying cell organization in multicellular differentiation, development, and disease progression.

bioinformatics↗

bioRxiv2023

PEFT-SP: Parameter-Efficient Fine-Tuning on Large Protein Language Models Improves Signal Peptide Prediction

Signal peptides (SP) play a crucial role in protein translocation in cells. The development of large protein language models (PLMs) provides a new opportunity for SP prediction, especially for the categories with limited annotated data. We present a Parameter-Efficient Fine-Tuning (PEFT) framework for SP prediction, PEFT-SP, to effectively utilize pre-trained PLMs. We implanted low-rank adaptation (LoRA) into ESM-2 models to better leverage the protein sequence evolutionary knowledge of PLMs. Experiments show that PEFT-SP using LoRA enhances state-of-the-art results, leading to a maximum MCC2 gain of 0.372 for SPs with small training samples and an overall MCC2 gain of 0.048. Furthermore, we also employed two other PEFT methods, i.e., Prompt Tunning and Adapter Tuning, into ESM-2 for SP prediction. More elaborate experiments show that PEFT-SP using Adapter Tuning can also improve the state-of-the-art results with up to 0.202 MCC2 gain for SPs with small training samples and an overall MCC2 gain of 0.030. LoRA requires fewer computing resources and less memory compared to Adapter, making it possible to adapt larger and more powerful protein models for SP prediction.

bioinformatics↗