1-20 of 4436
Image
Graphical abstract of the XenoBug pipeline showing ( A ) generation of diff...
Published: 01 May 2025
Figure 1.
Graphical abstract of the XenoBug pipeline showing ( A ) generation of different types of features and ( B ) using machine learning to predict pollutant-degrading enzymes and identify the source of the biodegrading enzymes.
Journal Article
EDITOR'S CHOICE
XenoBug: machine learning-based tool to predict pollutant-degrading enzymes from environmental metagenomes
NAR Genomics and Bioinformatics, Volume 7, Issue 2, June 2025, lqaf037, https://doi.org/10.1093/nargab/lqaf037
Published: 01 May 2025
Image
( A ) Ten-fold cross-validation performance of various problem transformati...
Published: 01 May 2025
Figure 3.
( A ) Ten-fold cross-validation performance of various problem transformation methods with ANN and RF models evaluated using accuracy and F1 score. ( B ) Ten-fold cross-validation performance of various problem transformation methods with ANN and RF models evaluated using hamming loss.
Image
Average 10-fold cross-validation accuracy, F1 score, and hamming loss for m...
Published: 01 May 2025
Figure 4.
Average 10-fold cross-validation accuracy, F1 score, and hamming loss for multiple problem transformation methods and the RF-based model for ( A ) EC 1 subclass classification, ( B ) EC 2 subclass classification, ( C ) EC 3 subclass classification, ( D ) EC 4 subclass classification, ( E ) EC 5 subc
Journal Article
GINClus: RNA structural motif clustering using graph isomorphism network
NAR Genomics and Bioinformatics, Volume 7, Issue 2, June 2025, lqaf050, https://doi.org/10.1093/nargab/lqaf050
Published: 26 April 2025
Journal Article
Discovering governing equations of biological systems through representation learning and sparse model discovery
NAR Genomics and Bioinformatics, Volume 7, Issue 2, June 2025, lqaf048, https://doi.org/10.1093/nargab/lqaf048
Published: 26 April 2025
Image
Graph representation of motif 1U9S_A:136-139_161-162. Here, A, G, and C rep...
Published: 26 April 2025
Figure 2.
Graph representation of motif 1U9S_A:136-139_161-162. Here, A, G, and C represent adenine, guanine and cytosine, respectively. ( A ) 2D representation of the base-pair interaction (solid line) and the base-stacking (dotted line) of the motif using the notations proposed by [ 28 ]. ( B ) Connected di
Image
( A ) Mean CCC by cell type between true proportions and estimated cell-typ...
Published: 26 April 2025
Figure 2.
( A ) Mean CCC by cell type between true proportions and estimated cell-type proportions from simulated pseudo-bulks. ( B ) CCC between estimated and true proportions of CT5 in the pseudo-bulks for the benchmark simulations. Each column represents a case of the relative abundance of CT5 against othe
Image
( A ) Box plots for estimated PBMC proportions by ARTdeConv in separate dec...
Published: 26 April 2025
Figure 5.
( A ) Box plots for estimated PBMC proportions by ARTdeConv in separate deconvolution analyses for healthy control and COVID-19-infected samples. ( B ) and ( C ) Box plots for estimated T cell and monocyte proportions on COVID-19-infected samples of different severity. ( D ) Scatter plots for estima
Image
Confusion matrix of GINClus model on test data for ( A ) internal loop moti...
Published: 26 April 2025
Figure 4.
Confusion matrix of GINClus model on test data for ( A ) internal loop motifs and ( B ) hairpin loop motifs.
Image
Experimental overview—evaluating sequence and structural similarity metrics...
Published: 26 April 2025
Figure 1.
Experimental overview—evaluating sequence and structural similarity metrics for predicting shared paralog functions. ( A ) A paralog pair A1–A2 is annotated based on their ability to perform shared functions. Two distinct labels are used: shared PPI (the two genes of the pair share a significative o
Image
Integrating all features improves prediction of shared PPIs, SL, and GO sem...
Published: 26 April 2025
Figure 6.
Integrating all features improves prediction of shared PPIs, SL, and GO semantic similarity. Performances of an XGBoost classifier using all 36 sequence similarity features together compared with a classifier using solely sequence identity, the classifier using the nine predicted structure similarit
Journal Article
EDITOR'S CHOICE
Evaluating sequence and structural similarity metrics for predicting shared paralog functions
NAR Genomics and Bioinformatics, Volume 7, Issue 2, June 2025, lqaf051, https://doi.org/10.1093/nargab/lqaf051
Published: 26 April 2025
Image
The overall pipeline used by GINClus. ( A ) Steps used to train the GIN mod...
Published: 26 April 2025
Figure 1.
The overall pipeline used by GINClus. ( A ) Steps used to train the GIN model. ( B ) Steps used to cluster the motif candidates (loop regions) using the trained GIN model.
Image
( A ) Mean absolute deviation (MAD) between true proportions and estimated ...
Published: 26 April 2025
Figure 3.
( A ) Mean absolute deviation (MAD) between true proportions and estimated cell-type proportions from simulated pseudo-bulks. ( B ) MAD between estimated and true proportions of CT5 in the pseudo-bulks for the benchmark simulations. Each column represents a case of the relative abundance of CT5 agai
Image
( A ) Estimated PBMC proportions by ARTdeConv versus true proportions measu...
Published: 26 April 2025
Figure 4.
( A ) Estimated PBMC proportions by ARTdeConv versus true proportions measured by flow cytometry for two PBMC samples collected on Day 0 with flexible mRNA amount parameters. ( B ) Estimated PBMC proportions by ARTdeConv versus true proportions measured by flow cytometry for two PBMC samples collect
Image
Panels ( A – D ) show the base-pair interaction (top) and 3D structure (bot...
Published: 26 April 2025
Figure 5.
Panels ( A – D ) show the base-pair interaction (top) and 3D structure (bottom) of example motifs from the families double-KT, 3-point-turn, anchor-loop, and bow-loop, respectively. Base-pair interaction notations used here are collected from [ 28 ]. Here, A, C, G, and U represent adenine, cytosine,
Image
CLERA discovers dynamical systems and gene programs from simulated data. ( ...
Published: 26 April 2025
Figure 1.
CLERA discovers dynamical systems and gene programs from simulated data. ( A ) Schematic of a two-gene regulatory network (G 1 and G 2 ) with discovered governing equations and parameters shown. ( B ) Comparison of generated gene expression data (top) and solutions from equations discovered by SIND
Image
CLERA uncovers dynamics and gene programs in pancreatic development. ( A ) ...
Published: 26 April 2025
Figure 2.
CLERA uncovers dynamics and gene programs in pancreatic development. ( A ) Discovered differential equations governing mouse pancreas development data from scRNA-seq, showing sparse and interpretable models and connections between latent variables. ( B ) Temporal dynamics of latent variables, which
