Biology subjects

Lee, N. K.

Publications and source records attributed to Lee, N. K..

8 recordsLinked to original sources

bioRxiv2023

Co-transcriptional translation amplifies mRNA noise in Escherichia coli

The variability in mRNA expression among isogenic cells exposed to identical environments is inherent. This variability originates from the inherent stochasticity of all processes underlying transcription. Although transcription and translation can occur simultaneously on the same mRNA molecule in bacteria, it is not well understood whether and how co-transcriptional translation contributes to variability in mRNA expression. Here, we studied the contribution of co-transcriptional translation to mRNA noise in E. coli cells. Using a transcription system physically decoupled from translation, we investigated the effect of ribosome binding to mRNA transcripts on variability in mRNA expression. We found that the propagation of RNAP noise to the mRNA level was increased by ribosome binding, leading to larger variations in the mRNA levels. We further demonstrated that ribosome binding increased the transcription initiation rate, resulting in the promoter becoming susceptible to RNAP noise. Co-transcriptional translation amplified transcriptional noise and modulated transcriptional bursting kinetics in bacterial cells.

biophysics↗

bioRxiv2022

Evolution-inspired augmentations improve deep learning for regulatory genomics

Deep neural networks (DNNs) hold promise for functional genomics prediction, but their generalization capability may be limited by the amount of available data. To address this, we propose EvoAug, a suite of evolution-inspired augmentations that enhance the training of genomic DNNs by increasing genetic variation. However, random transformation of DNA sequences can potentially alter their function in unknown ways. Thus, we employ a fine-tuning procedure using the original non-transformed data to preserve functional integrity. Our results demonstrate that EvoAug substantially improves the generalization and interpretability of established DNNs across prominent regulatory genomics prediction tasks, offering a robust solution for genomic DNNs.

bioinformatics↗

bioRxiv2021

Designing Interpretable Convolution-Based Hybrid Networks for Genomics

Hybrid networks that build upon convolutional layers with attention mechanisms have demon-strated improved performance relative to pure convolutional networks across many regulatory genome analysis tasks. Their inductive bias to learn long-range interactions provides an avenue to identify learned motif-motif interactions. For attention maps to be interpretable, the convolutional layer(s) must learn identifiable motifs. Here we systematically investigate the extent that architectural choices in convolution-based hybrid networks influence learned motif representations in first layer filters, as well as the reliability of their attribution maps generated by saliency analysis. We find that design principles previously identified in standard convolutional networks also generalize to hybrid networks. This work provides an avenue to narrow the spectrum of architectural choices when designing hybrid networks such that they are amenable to commonly used interpretability methods in genomics.

bioinformatics↗

bioRxiv2021

Regression QSAR Models for Predicting HIV-1 Integrase Inhibitors

The Human Immunodeficiency Virus (HIV) infection is a global pandemic that has claimed 33 million lives to date. One of the most efficacious treatment for naive or pre-treated HIV patients is with the HIV integrase strand transfer inhibitors (INSTIs). However, given that HIV treatment is life-long, the emergence of HIV-1 strains resistant to INSTIs is an imminent challenge. In this work, we showed two best regression QSAR models that were constructed using a boosted Random Forest algorithm [Formula] and a boosted K* algorithm [Formula] to predict the pIC50 values of INSTIs. Subsequently, the regression QSAR models were deployed against the Drugbank database for drug repositioning. The top ranked compounds were further evaluated for their target engagement activity using molecular docking studies and their potential as INSTIs evaluated from our literature search. Our study offers the first example of a large-scale regression QSAR modelling effort for discovering highly active INSTIs to combat HIV infection.

pharmacology and toxicology↗

bioRxiv2020

Predicting the impact of sequence motifs on gene regulation using single-cell data

BackgroundBinding of transcription factors (TFs) at proximal promoters and distal enhancers is central to gene regulation. Yet, identification of TF binding sites, also known as regulatory motifs, and quantification of their impact on gene expression remains challenging. ResultsHere we infer putative regulatory motifs along with their cell type-specific importance using a convolutional neural network trained on single-cell data. Comparison of the importance score to expression levels across cells allows us to identify the TFs most likely to be binding at a given motif. Using multiple mouse tissues we obtain a model with cell type resolution which explains 29% of the variance in gene expression. Finally, by applying scover to distal enhancers identified using scATAC-seq from the mouse cerebral cortex we characterize changes in distal regulatory motifs during development. ConclusionsIt is possible to identify regulatory motifs as well as their importance from single-cell data using a neural network model where all of the parameters and outputs are easily interpretable to the user.

bioinformatics↗

bioRxiv2019

A low-cost, rapidly integrated debubbler (RID) module for microfluidic cell culture applications

Microfluidic platforms use controlled fluid flow to provide physiologically relevant biochemical and biophysical cues to cultured cells in a well-defined and reproducible manner. In these systems, undisturbed flows are critical and air bubbles entering microfluidic channels can result in device delamination or cell damage. To prevent bubble entry, we report a low-cost, Rapidly Integrated Debubbler (RID) module that is simple to fabricate, inexpensive, and easily combined with existing experimental systems. We demonstrate successful removal of air bubbles spanning three orders of magnitude with a maximum removal rate (dV/dt)max = 1.5 mL min-1, at flow rates corresponding to physiological fluid-induced wall shear stresses (WSS) needed for biophysical stimulation studies on cultured mammalian cell populations.

bioengineering↗

bioRxiv2018

Clathrin-coated Pits are Shredded Organelles Presorting Receptors in the Plasma Membrane

A myriad of receptor types crowds the plasma membrane of cells. Here, we revealed that clathrin-coated pits (CCPs) presort receptors to modulate receptor activation on the plasma membrane. We visualized individual molecules of receptors, including epidermal growth factor receptor (EGFR), ErbB2, transferrin receptor (TfR), and beta2-adrenergic receptors ({beta}2-AR), inside a single CCP in a living cell using single-molecule diffusivity-based colocalization analysis. The spatially distinct subsets of CCPs selectively allocated for these receptors were observed in a resting state. The EGFR pre-allocated CCP subset was partially shared with that for ErbB2, whereas the EGFR and TfR pre-allocated CCP subsets were mutually exclusive. PICALM was necessary for the pre-allocation of CCPs for EGFR. Furthermore, EGFR dimerization was markedly elevated inside the pre-allocated CCP subset after dynamin recruitment. The pre-sorting function of CCPs provides an efficient mechanism to control competition and cooperation between receptors on the crowded plasma membrane.\n\nOne Sentence SummaryClathrin-coated pits are small-sized organelles pre-sorting receptor kinds to modulate receptor activation on the crowded plasma membrane.

cell biology↗

bioRxiv2017

Evaluation of Convolutionary Neural Networks Modeling of DNA Sequences using Ordinal versus one-hot Encoding Method

Convolutionary neural network (CNN) is a popular choice for supervised DNA motif prediction due to its excellent performances. To employ CNN, the input DNA sequences are required to be encoded as numerical values and represented as either vectors or multi-dimensional matrices. This paper evaluates a simple and more compact ordinal encoding method versus the popular one-hot encoding for DNA sequences. We compare the performances of both encoding methods using three sets of datasets enriched with DNA motifs. We found that the ordinal encoding performs comparable to the one-hot method but with significant reduction in training time. In addition, the one-hot encoding performances are rather consistent across various datasets but would require suitable CNN configuration to perform well. The ordinal encoding with matrix representation performs best in some of the evaluated datasets. This study implies that the performances of CNN for DNA motif discovery depends on the suitable design of the sequence encoding and representation. The good performances of the ordinal encoding method demonstrates that there are still rooms for improvement for the one-hot encoding method.

bioinformatics↗