Biology subjects

Baghdassarian, H. M.

Publications and source records attributed to Baghdassarian, H. M..

6 recordsLinked to original sources

bioRxiv2026

scLEMBAS: Context-Aware Modeling of Signaling Pathway Activity at Single-Cell Resolution

Cells sense and integrate extracellular cues through intracellular signaling networks that reshape transcription factor activity to dictate cellular responses. Signaling activity is difficult to decipher: it is non-linear, and it contains extensive feedback and crosstalk. Furthermore, the same perturbation can elicit markedly different responses depending on context (e.g., cell type, disease state, and tissue microenvironment) such that identical stimuli produce diverse responses in multicellular populations. Consequently, there is a vast combinatorial space of complex interactions and context-dependent responses that necessitate computational models. Computational models of single-cell perturbation responses are demonstrated to predict cellular responses, but are often limited in mechanistic insight. Prior knowledge networks offer a route to bridge predictive capability and interpretability. Here we present scLEMBAS, a context-aware, gray-box neural network that models signaling pathway activity at single-cell resolution while preserving mechanistic grounding. scLEMBAS encodes a prior-knowledge network of protein-protein interactions as a recurrent neural network whose learnable edge weights correspond to signaling interaction strengths. It also captures context and individual cell variance through compositional bias terms. An adversarial approach allows the model to answer a single-cell counterfactual - what a given cells TF activity would be under a different perturbation or context - while involving mechanistic rather than simply relational information. Across two scRNA-seq datasets spanning single- and multi-perturbation settings, scLEMBAS accurately predicts out-of-distribution combinations of perturbation and context. Capturing population variance across individual cells enables the model to predict cell subtype specific perturbation responses, despite being agnostic to such labels. Beyond prediction, scLEMBAS learned parameters are biologically interpretable: learned edge weights carry information beyond network topology and "self-prune" spurious interactions, while the categorical bias nominates proteins associated with cell-type-specific perturbation states. Overall, scLEMBAS enables quantitative dissection of how signaling pathway activity is reshaped by perturbation within specific cellular contexts.

bioinformatics↗

bioRxiv2026

Human Genome-Scale Models of Metabolism and Gene Expression Reveal Resource Constraints of Cancer Cell Lines

Genome-scale metabolic models (M-models) provide mechanistic insight into intracellular metabolism by simulating fluxes subject to nutrient and energy resource constraints. However, they cannot account for a major component of resource allocation, since they do not explicitly account for the cost of producing and maintaining enzymes. Genome-scale models of metabolism and gene expression (ME-Models) address this by including gene expression reactions, but these have only been developed for prokaryotes due to the additional complexity and challenges of modeling eukaryotes. Here, we present the human ME-Model, which encodes transcription, translation, complex formation, and turnover reactions for all enzymes catalyzing metabolic reactions, and couples these processes to constrain metabolic fluxes. We introduce humanME, a Python package to build and analyze human ME-Models. With this, we constructed 16 cancer cell line ME-Models. We found that resource constraints improve growth-rate predictions, and that ME-Model flux predictions are more biologically plausible and efficient. Moreover, transcriptional fluxes recapitulate RNA-Seq expression levels, with discrepancies revealing potential trade-offs involving multiple cellular objectives. Finally, the ME-Model recapitulates the Warburg effect, with increasing growth rate inducing glycolytic shifts, in part due to machinery costs of the electron transport chain. Altogether, we show ME-modeling can mechanistically link gene expression, resource allocation, and metabolism in human cells, substantially expanding the predictive scope of constraint-based models.

systems biology↗

bioRxiv2025

Joint linear modeling of transcriptomics and proteomics is predictive of cancer metastasis

A central goal of conducting omics measurements is to understand how molecular features inform higher-order cell- and tissue-level phenotypes. In particular, multi-omics offers insights into how information encoded by the genome is coordinated through biological layers, resulting in functional outputs1. Due to myriad post-transcriptional regulatory processes, the coordination between mRNA and protein cannot be simply reduced to gene-wise correlation. Yet, both modalities have been shown to serve as representations of biological state, and multi-omics integration has been used to improve these representations. Multi-omics approaches typically do not focus on how mRNA and protein features coordinate, but rather use the additional information for improved prediction or feature selection. Here, instead, we showed that standard linear machine learning models provide an understanding of transcriptomic and proteomic coordination in the context of a biological phenotype of interest, in this case cancer metastasis. We find that, in the context of metastasis, a select subset of proteomic features--reflecting a more concentrated signal relative to the broadly distributed transcriptomic signal--offers additional information to that encoded by transcriptomics, as demonstrated by improved model performance when integrating the two modalities and the relative feature importance of proteomics. Top features show a depletion of gene-product overlap across modalities, indicating that the model primarily leverages instances in which the two modalities are providing complementary information with respect to phenotype. However, in instances when both modalities are selected for a given gene product, there is high information consistency that synergistically bolsters phenotype prediction. Altogether, by using model fits that relate both modalities to phenotype, we observe a nuanced coordination of protein and mRNA, in which both modalities tend to provide consistent information about phenotype, yet benefits remain to incorporating a combination of both complementary and reinforcing signals across modalities.

systems biology↗

bioRxiv2024

Library-based single-cell analysis of CAR signaling reveals drivers of in vivo persistence

The anti-tumor function of engineered T cells expressing chimeric antigen receptors (CARs) is dependent on signals transduced through intracellular signaling domains (ICDs). Different ICDs are known to drive distinct phenotypes, but systematic investigations into how ICD architectures direct T cell function--particularly at the molecular level--are lacking. Here, we use single-cell sequencing to map diverse signaling inputs to transcriptional outputs, focusing on a defined library of clinically relevant ICD architectures. Informed by these observations, we functionally characterize transcriptionally distinct ICD variants across various contexts to build comprehensive maps from ICD composition to phenotypic output. We identify a unique tonic signaling signature associated with a subset of ICD architectures that drives durable in vivo persistence and efficacy in liquid, but not solid, tumors. Our findings work toward decoding CAR signaling design principles, with implications for the rational design of next-generation ICD architectures optimized for in vivo function.

immunology↗

bioRxiv2023

Combining LIANA and Tensor-cell2cell to decipher cell-cell communication across multiple samples

In recent years, data-driven inference of cell-cell communication has helped reveal coordinated biological processes across cell types. While multiple cell-cell communication tools exist, results are specific to the tool of choice, due to the diverse assumptions made across computational frameworks. Moreover, tools are often limited to analyzing single samples or to performing pairwise comparisons. As experimental design complexity and sample numbers continue to increase in single-cell datasets, so does the need for generalizable methods to decipher cell-cell communication in such scenarios. Here, we integrate two tools, LIANA and Tensor-cell2cell, which combined can deploy multiple existing methods and resources, to enable the robust and flexible identification of cell-cell communication programs across multiple samples. In this protocol, we show how the integration of our tools facilitates the choice of method to infer cell-cell communication and subsequently perform an unsupervised deconvolution to obtain and summarize biological insights. We explain how to perform the analysis step-by-step in both Python and R, and we provide online tutorials with detailed instructions available at https://ccc-protocols.readthedocs.io/. This protocol typically takes [~]1.5h to complete from installation to downstream visualizations on a GPU-enabled computer, for a dataset of [~]63k cells, 10 cell types, and 12 samples.

bioinformatics↗

bioRxiv2020

Multiple freeze-thaw cycles lead to a loss of consistency in poly(A)-enriched RNA sequencing

RNA-Seq is ubiquitous, but depending on the study, sub-optimal sample handling may be required, resulting in repeated freeze-thaw cycles. However, little is known about how each cycle impacts downstream analyses, due to a lack of study and known limitations in common RNA quality metrics, e.g., RIN, at quantifying RNA degradation following repeated freeze-thaws. Here we quantify the impact of repeated freeze-thaw on the reliability of downstream RNA-Seq analysis. To do so, we developed a method to estimate the relative noise between technical replicates independently of RIN. Using this approach we inferred the effect of both RIN and the number of freeze-thaw cycles on sample noise. We find that RIN is unable to fully account for the change in sample noise due to freeze-thaw cycles. Additionally, freeze-thaw is detrimental to sample quality and differential expression (DE) reproducibility, approaching zero after three cycles for poly(A)-enriched samples, wherein the inherent 3 bias in read coverage is more exacerbated by freeze-thaw cycles, while ribosome-depleted samples are less affected by freeze-thaws. The use of poly(A)-enrichment for RNA sequencing is pervasive in library preparation of frozen tissue, and thus, it is important during experimental design and data analysis to consider the impact of repeated freeze-thaw cycles on reproducibility. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=103 SRC="FIGDIR/small/020792v2_ufig1.gif" ALT="Figure 1"> View larger version (19K): org.highwire.dtl.DTLVardef@c2c756org.highwire.dtl.DTLVardef@1ad1c03org.highwire.dtl.DTLVardef@a4136org.highwire.dtl.DTLVardef@13f403f_HPS_FORMAT_FIGEXP M_FIG C_FIG

bioinformatics↗