Within the Natural Language Processing (NLP) framework, Named Entity Recognition (NER) is regarded as the basis for extracting key information to understand texts in any language. As Bangla is a highly inflectional, morphologically rich, and resource-scarce language, building a balanced NER corpus with large and diverse entities is a demanding task. However, previously developed Bangla NER systems are limited to recognizing only three familiar entities: person, location, and organization. To address this significant limitation, we introduce a novel Bangla NER dataset B-NER, which was created using 22,144 manually annotated Bangla sentences collected from Bangla newspapers and Bangla Wikipedia. This dataset includes a total of 9,895 unique words which were manually categorized into eight different entity types, such as a person, organization, event, artifact, time indicator, natural phenomenon, geopolitical entity, and geographical location. Inter-annotator agreement experiments were conducted to validate the quality of annotations performed by three annotators, resulting in a Kappa score of 0.82. In this paper, we provide an outline of the annotation guideline illustrated with examples, discuss the B-NER dataset properties, and present benchmark evaluations of the dataset. To establish that B-NER is more comprehensive and balanced in comparison to other publicly accessible datasets, we conducted cross-dataset modeling and validation, i.e. trained NER model on one dataset while tested on another, and found that the model trained on B-NER performed the best in that settings. Furthermore, we performed exhaustive benchmark evaluations based on Bidirectional LSTM with fastText embeddings and sentence transformer models. Among these models, fine-tuned IndicBERT achieved noticeable results with a Macro-F1 of 86%. This dataset and baseline results will be publicly available under a CC-BY 4.0 license in the CoNLL-2002 format to facilitate further research on Bangla NER.

We introduce a protein language model for determining the complete sequence of a peptide based on measurement of a limited set of amino acids. To date, protein sequencing relies on mass spectrometry, with some novel Edman degradation based platforms able to sequence non-native peptides. Current protein sequencing techniques face limitations in accurately identifying all amino acids, hindering comprehensive proteome analysis. Our method simulates partial sequencing data by selectively masking amino acids that are experimentally difficult to identify in protein sequences from the UniRef database. This targeted masking mimics real-world sequencing limitations. We then modify and finetune a ProtBERT derived transformer-based model, for a new downstream task predicting these masked residues, providing an approximation of the complete sequence. Evaluating on three bacterial Escherichia species, we achieve per-amino-acid accuracy up to 90.5% when only four amino acids ([KCYM]) are known. Structural assessment using AlphaFold2 and TM-score validates the biological relevance of our predictions. The model also demonstrates potential for evolutionary analysis through cross-species performance. This integration of simulated experimental constraints with computational predictions offers a promising avenue for enhancing protein sequence analysis, potentially accelerating advancements in proteomics and structural biology by providing a probabilistic reconstruction of the complete protein sequence from limited experimental data.

In this study, we apply machine learning to model the spatiotemporal dynamics of gene expression during early Drosophila embryogenesis. By optimizing model architecture, feature selection, and spatial grid resolution, we developed a predictive pipeline capable of accurately classifying active nuclei and forecasting their future distribution in time. We evaluated the model on two reporter constructs for the short gastrulation (sog) gene, sogD and sogD_∆Su(H), allowing us to assess its performance across distinct genetic contexts. The model achieved high accuracy on the wild-type sogD dataset, particularly along the dorsal–ventral (DV) axis during nuclear cycle 14 (NC14), and accurately predicted expression in the central regions of both wild-type and Suppressor of Hairless (Su(H)) mutant enhancers, sogD_∆Su(H). Bootstrap analysis confirmed that the model performed better in the central region than at the edges, where prediction accuracy dropped. Our previous work showed that Su(H) can act both as a repressor at the borders and as a stabilizer of transcriptional bursts in the center of the sog expression domain. This dual function is not unique to Su(H); other broadly expressed transcription factors also exhibit context-dependent regulatory roles, functioning as activators in some regions and repressors in others. These results highlight the importance of spatial context in transcriptional regulation and demonstrate the ability of machine learning to capture such nuanced behavior. Looking ahead, incorporating mechanistic features such as transcriptional bursting parameters into predictive models could enable simulations that forecast not just where genes are expressed but also how their dynamics unfold over time. This form of in silico enhancer mutagenesis would make it possible to predict the effects of specific binding site changes on both spatial expression patterns and underlying transcriptional activity, offering a powerful framework for studying cis-regulatory logic and modeling early developmental processes across diverse genetic backgrounds.