Biomedical subjects

Xiaofang Jiang

Publications and source records attributed to Xiaofang Jiang.

5 recordsLinked to original sources

PubMed2026

LAMBDA: a prophage detection benchmark for genomic language models.

Transformer-based genomic sequence models represent an emerging frontier in computational biology. Yet, their embeddings have not yet shown the same level of predictive power as natural and protein language models, highlighting a gap between current implementations and theoretical promise. Existing benchmarks for DNA language models primarily focus on classifying regulatory elements in eukaryotic genomes, leaving open the fundamental question of whether these models learn sequence-level features across whole genomes. We introduce LAMBDA, a benchmark designed to rigorously evaluate genome language model embeddings through phage-bacteria sequence discrimination across four categories of increasing complexity: probing tasks, fine-tuning assessments, diagnostic tests, and genome-wide prophage detection. Our comprehensive analysis of current genomic language models provides insight into the importance of training data selection relative to model size, the need for domain-specific training, and the capabilities and limitations of genomic language models for detecting prophage sequences. This benchmark represents a challenging genomic annotation task in the bacterial domain and addresses a key computational problem with direct relevance to microbiology and medicine.

Prophages↗

PubMed2026

LAMBDA: A Prophage Detection Benchmark for Genomic Language Models.

Transformer-based genomic sequence models represent an emerging frontier in computational biology. Yet, their embeddings have not yet shown the same level of predictive power as natural and protein language models, indicating a gap between current implementations and theoretical promise. Existing benchmarks for DNA language models primarily focus on classifying regulatory elements in eukaryotic genomes, leaving open the fundamental question of whether these models learn sequence-level features across whole genomes. We introduce LAMBDA, a benchmark designed to rigorously evaluate genome language model embeddings through phage-bacteria sequence discrimination across four categories of increasing complexity: probing tasks, fine-tuning assessments, diagnostic tests, and genome-wide prophage detection. Our comprehensive analysis of current genomic language models provides novel insights into the importance of training data quality relative to model size, the need for domain-specific training, and the application of genomic language models for detecting prophage sequences. This benchmark represents a challenging genomic annotation task in the bacterial domain and addresses a key computational problem with direct relevance to microbiology and medicine.

DNA language model↗

PubMed2025

The impact of tokenizer selection in genomic language models.

MOTIVATION: Genomic language models have recently emerged as a new method to decode, interpret, and generate genetic sequences. Existing genomic language models have utilized various tokenization methods, including character tokenization, overlapping and nonoverlapping k-mer tokenization, and byte-pair encoding, a method widely used in natural language models. Genomic sequences differ from natural language because of their low character variability, complex and overlapping features, and inconsistent directionality. These features make subword tokenization in genomic language models significantly different from both traditional language models and protein language models. RESULTS: This study explores the impact of tokenization in genomic language models by evaluating their downstream performance on 44 classification fine-tuning tasks. We also perform a direct comparison of byte pair encoding and character tokenization in Mamba, a state-space model. Our results indicate that character tokenization outperforms subword tokenization methods on tasks that rely on nucleotide-level resolution, such as splice site prediction and promoter detection. While byte-pair tokenization had stronger performance on the SARS-CoV-2 variant classification task, we observed limited statistically significant differences between tokenization methods on the remaining downstream tasks. AVAILABILITY AND IMPLEMENTATION: Detailed results of all benchmarking experiments are available in https://github.com/leannmlindsey/DNAtokenization. Training datasets and pretrained models are available at https://huggingface.co/datasets/leannmlindsey. Datasets and processing scripts are available at doi: 10.5281/zenodo.16287401 and doi: 10.5281/zenodo.16287130.

Natural Language Processing↗

PubMed2006

Cryoelectron microscopy structures of rotavirus NSP2-NSP5 and NSP2-RNA complexes: implications for genome replication.

The replication and packaging of the rotavirus genome, comprising 11 segments of double-stranded RNA, take place in specialized compartments called viroplasms, which are formed during infection and involve a coordinated interplay of multiple components. Two rotavirus nonstructural proteins, NSP2 (with nucleoside triphosphatase, single-stranded RNA [ssRNA] binding and helix-destabilizing activities) and NSP5, are essential in these events. Previous structural analysis of NSP2 showed that it is an octamer in crystals, obeying 4-2-2 crystal symmetry, with a large 35-A central hole along the fourfold axis and deep grooves at one of the twofold axes. To ascertain that the solution structure of NSP2 is the same as that in the crystals and investigate how NSP2 interacts with NSP5 and RNA, we carried out single-particle cryoelectron microscopy (cryo-EM) analysis of NSP2 alone and in complexes with NSP5 and ssRNA at subnanometer resolution. Because full-length NSP5 caused severe aggregation upon mixing with NSP2, the deletion construct NSP566-188 was used in these studies. Our studies show that the solution structure of NSP2 is same as the crystallographic octamer and that both NSP566-188 and ssRNA bind to the grooves in the octamer, which are lined by positively charged residues. The fitting of the NSP2 crystal structure to cryo-EM reconstructions of the complexes indicates that, in contrast to the binding of NSP566-188, the binding of RNA induces noticeable conformational changes in the NSP2 octamer. Consistent with the observation that both NSP5 and RNA share the same binding site on the NSP2 octamer, filter binding assays showed that NSP5 competes with ssRNA binding, indicating that one of the functions of NSP5 is to regulate NSP2-RNA interactions during genome replication.

Binding Sites↗

PubMed2006

Structure-function analysis of rotavirus NSP2 octamer by using a novel complementation system.

Viral inclusion bodies, or viroplasms, that form in rotavirus-infected cells direct replication and packaging of the segmented double-stranded RNA (dsRNA) genome. NSP2, one of two rotavirus proteins needed for viroplasm assembly, possesses NTPase, RNA-binding, and helix-unwinding activities. NSP2 of the rotavirus group causing endemic infantile diarrhea (group A) was shown to self-assemble into large doughnut-shaped octamers with circumferential grooves and deep clefts containing nucleotide-binding histidine triad (HIT)-like motifs. Here, we demonstrate that NSP2 of group C rotavirus, a group that fails to reassort with group A viruses, retains the unique architecture of the group A octamer but differs in surface charge distribution. By using an NSP2-dependent complementation system, we show that the HIT-dependent NTPase activity of NSP2 is necessary for dsRNA synthesis, but not for viroplasm formation. The complementation system also showed that despite the retention of the octamer structure and the HIT-like fold, group C NSP2 failed to rescue replication and viroplasm formation in NSP2-deficient cells infected with group A rotavirus. The distinct differences in the surface charges on the Bristol and SA11 NSP2 octamers suggest that charge complementarity of the viroplasm-forming proteins guides the specificity of viroplasm formation and, possibly, reassortment restriction between rotavirus groups.

Amino Acid Sequence↗