SIGRN: Inferring Gene Regulatory Network with Soft Introspective Variational Autoencoders

International Journal of Molecular Sciences, Год журнала: 2024, Номер 25(23), С. 12741 - 12741

Опубликована: Ноя. 27, 2024

Gene regulatory networks (GRNs) exhibit the complex relationships among genes, which are essential for understanding developmental biology and uncovering fundamental aspects of various biological phenomena. It is an effective economical way to infer GRNs from single-cell RNA sequencing (scRNA-seq) with computational methods. Recent researches have been done on problem by using variational autoencoder (VAE) structural equation model (SEM). Due shortcoming VAE generating poor-quality data, in this paper, a soft introspective adversarial gene network unsupervised inference model, called SIGRN, proposed introducing mechanism building model. SIGRN applies “soft” mode avoid training additional neural adding parameters. demonstrates superior accuracy across most benchmark datasets when compared nine leading-edge In addition, method also achieves better performance representing cells scRNA-seq data datasets. All verified via substantial experiments. The shows promise inferring GRNs.

Язык: Английский

---

Topological radiogenomics based on persistent lifetime images for identification of epidermal growth factor receptor mutation in patients with non-small cell lung tumors

Computers in Biology and Medicine, Год журнала: 2024, Номер 185, С. 109519 - 109519

Опубликована: Дек. 11, 2024

Язык: Английский

---

Variations towards an efficient drug–drug interaction

The Computer Journal, Год журнала: 2024, Номер unknown

Опубликована: Дек. 11, 2024

Abstract Drug–drug interactions (DDIs) are a crucial research focus in clinical pharmacology and public health. DDIs can lead to reduced drug efficacy or increased adverse reactions, making the effective identification understanding of essential for patient safety treatment outcomes. With rapid growth biomedical literature, automated methods extracting DDI information have become increasingly necessary. In this paper, we propose BLRG, novel model that uniquely integrates BioBERT, long short-term memory (LSTM), relational graph convolutional network (R-GCN) extract complex DDIs. This combination allows effectively capture both semantic features, outperforming existing handling intricate dependencies texts. Specifically, our approach begins by utilizing BioBERT deep contextual features sentences, their information. Following this, an LSTM processes sequential sentence its dependencies. Finally, R-GCN is applied identify interpret relationships between entities within sentence, accurately capturing Experimental results demonstrate significantly outperforms current state-of-the-art across standard datasets, showcasing effectiveness potential extraction tasks. Our code data publicly available at: https://github.com/Hero-Legend/BLRG.

Язык: Английский

---

Rethinking cancer drug synergy prediction: a call for standardization in machine learning applications

bioRxiv (Cold Spring Harbor Laboratory), Год журнала: 2024, Номер unknown

Опубликована: Дек. 24, 2024

Abstract Drug resistance poses a significant challenge to cancer treatment, often caused by intratumor heterogeneity. Combination therapies have been shown be an effective strategy prevent resistant cells from escaping single-drug treatments. However, discovering new drug combinations through traditional molecular assays can costly and time-consuming. In silico approaches overcome this limitation exploring many candidate at scale. This study systematically evaluates the utility of various machine learning algorithms, input features, synergy prediction tasks. Our findings indicate pressing need for establishing standardized framework measure develop algorithms capable predicting synergy.

Язык: Английский

---

Learning genotype–phenotype associations from gaps in multi-species sequence alignments

Briefings in Bioinformatics, Год журнала: 2024, Номер 26(1)

Опубликована: Ноя. 22, 2024

Abstract Understanding the genetic basis of phenotypic variation is fundamental to biology. Here we introduce GAP, a novel machine learning framework for predicting binary phenotypes from gaps in multi-species sequence alignments. GAP employs neural network predict presence or absence solely alignment gaps, contrasting with existing tools that require additional and often inaccessible input data. can be applied three distinct problems: species known associated genomic regions, pinpointing positions within such regions are important phenotypes, extracting sets candidate phenotypes. We showcase utility by exploiting well-known association between L-gulonolactone oxidase (Gulo) gene vitamin C synthesis, demonstrating its perfect prediction accuracy 34 vertebrates. This exceptional performance also applies more generally, achieving high power on large simulated dataset. Moreover, predictions synthesis unknown status mirror their phylogenetic relationships, predictive importance consistent those identified previous studies. Last, genome-wide application identifies many genes may analysis these candidates uncovers functional enrichment immunity, widely recognized role C. Hence, represents simple yet useful tool genotype–phenotype associations addressing diverse evolutionary questions data available broad range study systems.

Язык: Английский

---