Biology subjects

Phung, T.

Publications and source records attributed to Phung, T..

5 recordsLinked to original sources

bioRxiv2025

XpressO-Melanoma: An Explainable Deep Learning Model for the Prediction of BRAF V600 Mutation Status in Cutaneous Melanomas

BRAF V600E mutations are critical oncogenic drivers in cutaneous melanoma, influencing treatment decisions and outcomes. However, conventional molecular assays face limitations, including tissue availability, cost, and access. To address this, we present an explainable deep learning model that predicts BRAF V600E mutation status directly from diagnostic whole-slide images (WSIs) of skin cutaneous melanoma. Using histopathological WSIs from The Cancer Genome Atlas (TCGA) and their corresponding mutation labels (BRAF wildtype vs. BRAF V600E), we trained a weakly supervised deep learning pipeline, XpressO, to identify tumor regions of interest (ROIs) predictive of BRAF mutation status. The model outputs attention heatmaps highlighting spatially relevant diagnostic features and computes a combined probability score from the top ten attention regions per WSI. These regions are further reviewed by a pathologist for biological appropriateness. On an independent test set, the model achieved an AUC of 0.79 with balanced precision and recall, correctly identifying 7 of 8 BRAF V600E mutant cases. This demonstrates the models ability to capture phenotypic correlates of mutation status and highlights the potential of computational pathology in precision oncology. Our approach offers a scalable, interpretable, and cost-effective alternative to molecular testing, particularly in resource-limited settings.

cancer biology↗

bioRxiv2025

Data-independent acquisition (DIA) approach for comprehensive ubiquitinome profiling in targeted protein degradation

Targeted protein degradation (TPD) has emerged as a highly promising therapeutic strategy for a wide range of diseases, including cancer and neurodegenerative disorders. The ubiquitin-proteasome system, which is responsible for protein degradation, plays a critical role in this process. Gaining comprehensive insights into the ubiquitylation landscape is essential for the development of selective and efficient targeted protein degradation approaches. Recently, data-independent acquisition (DIA) has gained significant popularity as a robust and unbiased approach for quantitative proteomics. Here, we report a cutting-edge workflow that utilizes diGly antibody-based enrichment followed by an optimized Orbitrap-based DIA method for the identification of ubiquitylated peptides. We identify over 40,000 diGly precursors corresponding to more than 7,000 proteins in a single measurement from cells exposed to a proteasome inhibitor, highlighting an exceptional throughput. By applying our optimized workflow, we successfully identify ubiquitylation sites on substrate proteins with various TPD approaches. Therefore, our workflow holds tremendous potential for rapidly establishing mode of action for various TPD modalities, including PROTACs and molecular glues.

cell biology↗

bioRxiv2023

Adaptations to water stress and pastoralism in the Turkana of northwest Kenya

The Turkana people inhabit arid regions of east Africa--where temperatures are high and water is scarce--and they practice subsistence pastoralism, such that their diet is primarily composed of animal products. Working with Turkana communities, we sequenced 367 genomes and identified 8 regions putatively involved in adaptation to water stress and pastoralism. One of these regions includes a putative enhancer for STC1--a kidney-expressed gene involved in the response to dehydration and the metabolism of purine-rich foods such as red meat. We show that STC1 is induced by antidiuretic hormone in humans, is associated with urea levels in the Turkana themselves, and is under strong selection in this population (s[~]0.041). This work highlights that partnerships with subsistence-level groups can lead to new models of human physiology with biomedical relevance.

evolutionary biology↗

bioRxiv2018

Identifying, understanding, and correcting technical biases on the sex chromosomes in next-generation sequencing data

Mammalian X and Y chromosomes share a common evolutionary origin and retain regions of high sequence similarity. This sequence homology can cause the mismapping of short sequencing reads derived from the sex chromosomes and affect variant calling and other downstream analyses. Understanding and correcting this problem is critical for medical genomics and population genomic inference. Here, we characterize how sequence homology can affect analyses on the sex chromosomes and present XYalign, a new tool that: (1) facilitates the inference of sex chromosome complement from next-generation sequencing data; (2) corrects erroneous read mapping on the sex chromosomes; and (3) tabulates and visualizes important metrics for quality control such as mapping quality, sequencing depth, and allele balance. We show how these metrics can be used to identify XX and XY individuals across diverse sequencing experiments, including low and high coverage whole genome sequencing, and exome sequencing. We also show that XYalign corrects mismapped reads on the sex chromosomes, resulting in more accurate variant calling. Finally, we discuss how the flexibility of the XYalign framework can be leveraged for other use cases including the identification of aneuploidy on the autosomes. XYalign is available open source under the GNU General Public License (version 3).

bioinformatics↗

bioRxiv2017

Comparison of single genome and allele frequency data reveals discordant demographic histories

Inference of demographic history from genetic data is a primary goal of population genetics of model and non-model organisms. Whole genome-based approaches such as the Pairwise/Multiple Sequentially Markovian Coalescent (PSMC/MSMC) methods use genomic data from one to four individuals to infer the demographic history of an entire population, while site frequency spectrum (SFS)-based methods use the distribution of allele frequencies in a sample to reconstruct the same historical events. Although both methods are extensively used in empirical studies and perform well on data simulated under simple models, there have been only limited comparisons of them in more complex and realistic settings. Here we use published demographic models based on data from three human populations (Yoruba (YRI), descendants of northwest-Europeans (CEU), and Han Chinese (CHB)) as an empirical test case to study the behavior of both inference procedures. We find that several of the demographic histories inferred by the whole genome-based methods do not predict the genome-wide distribution of heterozygosity nor do they predict the empirical SFS. However, using simulated data, we also find that the whole genome methods can reconstruct the complex demographic models inferred by SFS-based methods, suggesting that the discordant patterns of genetic variation are not attributable to a lack of statistical power, but may reflect unmodeled complexities in the underlying demography. More generally, our findings indicate that demographic inference from a small number of genomes, routine in genomic studies of nonmodel organisms, should be interpreted cautiously, as these models cannot recapitulate other summaries of the data.

evolutionary biology↗