Large language models (LLMs) have emerged as powerful tools for biological sequence analysis. However, their applicability to the transcriptional regulation of long non-coding RNAs (IncRNAs) remains underexplored due to the complexity and diversity of IncRNA sequences, combined with limited knowledge of their regulatory mechanisms and functional characteristics. In this study, we systematically evaluate both singletask and multi-task fine-tuning strategies of genome foundation models across four tasks designed to capture increasing biological complexity. By fine-tuning genome foundation models on a series of progressively complex tasks, each designed to closely mimic the complexities of IncRNA classification, we explore how task complexity impacts model performance and biological interpretability. Our findings reveal that while foundation models capture promoter-specific signals, task complexity, input length, and data quality significantly influence performance. Notably, the most challenging task, promoter sequence classification of IncRNA from coding genes, remains limited by primary sequence features alone. Through performance evaluation and attentionbased interpretability analysis, this work provides key insights into the potential and limitations of current LLM-based models and offers a foundation for future development of integrative models in IncRNA regulation analysis. The dataset is available at https://github.com/wangwe90/IncRNA-promoter-sequence-detection-dataset

Identifying the cell populations present in a single cell RNAseq (sc-RNAseq) dataset is made difficult by the high-dimensionality and sparse aspects of the data. Often the first step at resolving this challenge is to perform feature selection; selecting a set of informative genes in the dataset to use in cell embedding and clustering. The typical sc-RNAseq feature selection methods choose a subset of genes with largest variances in their detected expressions across single cells. Here we show these conventional feature selection methods are susceptible to inflated variances due to inconsistent transcriptomic sampling. As an alternative, we present a computational algorithm that uses the binary correlations (co-occurrences) between genes to perform feature selection. Using multiple sc-RNAseq datasets, we show this co-occurrence based feature selection approach outperforms popular high-variance feature selection methods in terms of cell clustering accuracy and separability. Taken together, these results suggest that the co-occurrence based method may be more appropriate for performing feature selection in sc-RNAseq data, and it can be easily implemented for most sc-RNAseq workflows. Additional details of the co-occurrence feature selection algoirthm and supplementary materials are available at https://github.com/ncsu-penglab/cooccur_feature_selection [1].

A few studies suggest that many protein coding loci also transcribe extra-coding RNAs (ecRNAs), a distinct class of long non-coding RNAs. These ecRNAs span over and beyond mRNAs originating from the same loci, are much less abundant than mRNAs, and lack polyadenylation. Previously, ecRNAs were found to regulate the transcription of mRNAs from the same loci epigenetically. However, further investigation of ecRNAs is impeded by the lack of accurate ecRNA annotation and quantification. Here, to uncover these mRNA-overlapping ecRNAs, we first depleted polyadenylated RNAs, i.e. mRNAs, from total RNAs and performed deep sequencing analysis of the remaining total RNAs. Next, we devised a computational strategy to identify ecRNA transcribing loci from this unique sequencing data. Our approach uses a combination of features including the ratio of intronic to exonic read counts, the percent read coverage of introns, exons, and flanking regions, and the ratio of flanking to intronic read counts. In this pilot study of HIV-1 infected CD4+ T cells, we observed transcriptional signatures of ecRNAs from 880 (~11%) of 7,690 expressed coding genes. We expect these custom computational developments will enable a better understanding of the involvement of ecRNAs in regulating host response to HIV infection and in other conditions.

MOTIVATION: Quantification of sequence abundance in RNA-Seq experiments is often conflated by protocol-specific sequence bias. The exact sources of the bias are unknown, but may be influenced by polymerase chain reaction amplification, or differing primer affinities and mixtures, for example. The result is decreased accuracy in many applications, such as de novo gene annotation and transcript quantification. RESULTS: We present a new method to measure and correct for these influences using a simple graphical model. Our model does not rely on existing gene annotations, and model selection is performed automatically making it applicable with few assumptions. We evaluate our method on several datasets, and by multiple criteria, demonstrating that it effectively decreases bias and increases uniformity. Additionally, we provide theoretical and empirical results showing that the method is unlikely to have any effect on unbiased data, suggesting it can be applied with little risk of spurious adjustment. AVAILABILITY: The method is implemented in the seqbias R/Bioconductor package, available freely under the LGPL license from http://bioconductor.org CONTACT: [email protected] SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

MOTIVATION: The sequencing of the Plasmodium yoelii genome, a model rodent malaria parasite, has greatly facilitated research for the development of new drug and vaccine candidates against malaria. Unfortunately, only preliminary gene models were annotated on the partially sequenced genome, mostly by in silico gene prediction, and there has been no major improvement of the annotation since 2002. RESULTS: Here we report on a systematic assessment of the accuracy of the genome annotation based on a detailed analysis of a comprehensive set of cDNA sequences and proteomics data. We found that the coverage of the current annotation tends to be biased toward genes expressed in the blood stages of the parasite life cycle. Based on our proteomic analysis, we estimate that about 15% of the liver stage proteome data we have generated is absent from the current annotation. Through comparative analysis we identified and manually curated a further 510 P. yoelii genes which have clear orthologs in the P. falciparum genome, but were not present or incorrectly annotated in the current annotation. This study suggests that improvements of the current P. yoelii genome annotation should focus on genes expressed in stages other than blood stages. Comparative analysis will be critically helpful for this re-annotation. The addition of newly annotated genes will facilitate the use of P. yoelii as a model system for studying human malaria. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

The analysis of proteome profiles offers a new approach to understanding how cellular machinery functions and responds under certain conditions. By combining two-dimensional electrophoresis with mass spectrometry (MS), a snapshot of the cell's protein expression status and quantitative proteome profiling can be provided. As the cell's proteome becomes defined in normal and altered states, possible utilization of MS proteome profiling as a diagnostic tool becomes a reality. The ability of Matrix Assisted Laser Desorption Ionization Mass Spectrometry (MALDI-MS) to generate a spectrum with thousands of data points, necessitate the development of sophisticated analytical algorithms. In this paper, we describe how MALDI-MS can be used in monitoring proteomic profile in patients before and after treatment using a non- invasive sampling method. Because data analysis in this process possesses a challenge, we present a novel mathematical approach for analyzing data produced by MALDI MS, and discuss current applications of mass spectrometry in clinical medicine as well as challenges faced during procedures and experimentation. As a case study, we analyze protein expression patterns in premenopausal versus postmenopausal women. We also provide a proteomic profiling of premenopausal women versus postmenopausal women treated with estrogen as a hormone replacement therapy. * Corresponding authors:

BACKGROUND: The analysis of biological data is greatly enhanced by existing or emerging databases. Most existing databases, with few exceptions are not designed to easily support large scale computational analysis, but rather offer exclusively a web interface to the resource. We have recognized the growing need for a database which can be used successfully as a backend to computational analysis tools and pipelines. Such database should be sufficiently versatile to allow easy system integration. RESULTS: GeneKeyDB is a gene-centered relational database developed to enhance data mining in biological data sets. The system provides an underlying data layer for computational analysis tools and visualization tools. GeneKeyDB relies primarily on existing database identifiers derived from community databases (NCBI, GO, Ensembl, et al.) as well as the known relationships among those identifiers. It is a lightweight, portable, and extensible platform for integration with computational tools and analysis environments. CONCLUSION: GeneKeyDB can enable analysis tools and users to manipulate the intersections, unions, and differences among different data sets.