Biology subjects

Behsaz, B.

Publications and source records attributed to Behsaz, B..

4 recordsLinked to original sources

bioRxiv2025

A multi-modal cell-free RNA language model for liquid biopsy applications

Cell-free RNA (cfRNA) profiling has emerged as a powerful tool for non-invasive disease detection, but its application is limited by data sparsity and complexity, especially in settings with constrained sample availability. We introduce Exai-1, a multi-modal, transformer-based generative foundation model that integrates RNA sequence embeddings with cfRNA abundance data to capture biologically meaningful representations of circulating RNAs. By leveraging both sequence and expression modalities, Exai-1 captures a biologically meaningful latent structure of cfRNA profiles. Pre-trained on over 306 billion tokens from 8,339 samples, Exai-1 enhances signal fidelity, reduces technical noise, and improves disease detection by generating synthetic cfRNA profiles. We show that self-attention and variational inference are particularly important for preservation of biological signals and contextual relationships. Additionally, Exai-1 facilitates cross-biofluid translation and assay compatibility through disentangling biological signals from confounders. By uniting sequence-informed embeddings with cfRNA expression patterns, Exai-1 establishes a transfer learning foundation for liquid biopsy, offering a scalable and adaptable framework for next-generation cfRNA-based diagnostics.

bioinformatics↗

bioRxiv2025

Interpretable adenylation domain specificity prediction using protein language models

Natural products have long been a rich source of diverse and clinically effective drug candidates. Non-ribosomal peptides (NRPs), polyketides (PKs), and NRP-PK hybrids are three classes of natural products that display a broad range of bioactivities, including antibiotic, antifungal, anticancer, and immunosuppressant activities. However, discovering these compounds through traditional bioactivity-guided techniques is costly and time-consuming, often resulting in the rediscovery of known molecules. Consequently, genome mining has emerged as a high-throughput strategy to screen hundreds of thousands of microbial genomes to identify their potential to produce novel natural products. Adenylation domains play a key role in the biosynthesis of NRPs and NRP-PKs by recruiting substrates to incrementally build the final structure. We propose MASPR, a machine learning method that leverages protein language models for accurate and interpretable predictions of A-domain substrate specificities. MASPR demonstrates superior accuracy and generalization over existing methods and is capable of predicting substrates not present in its training data, or zero-shot classification. We use MASPR to develop Seq2Hybrid, an efficient algorithm to predict the structure of hybrid NRP-PK natural products from microbial genomes. Using Seq2Hybrid, we propose putative biosynthetic gene clusters for the orphan natural products Octaminomycin A, Dityromycin, SW-163B, and JBIR-39.

bioinformatics↗

bioRxiv2019

De Novo Peptide Sequencing Reveals a Vast Cyclopeptidome in Human Gut and Other Environments

Cyclic and branch cyclic peptides (cyclopeptides) represent an important class of bioactive natural products that include many antibiotics and anti-tumor compounds. However, little is known about cyclopeptides in the human gut, despite the fact that humans are constantly exposed to them. To address this bottleneck, we developed the CycloNovo algorithm for de novo cyclopeptide sequencing that employs de Bruijn graphs, the workhorse of DNA sequencing algorithms. CycloNovo reconstructed many new cyclopeptides that we validated with transcriptome, metagenome, and genome mining analyses. Our benchmarking revealed a vast hidden cyclopeptidome in the human gut and other environments and suggested that CycloNovo offers a much-needed step-change for cyclopeptide discovery. Furthermore, CycloNovo revealed a wealth of anti-microbial cyclopeptides from food that survive the complete human gastrointestinal tract, raising the question of how these cyclopeptides might affect the human microbiome.\n\nSIGNIFICANCEThe golden age of antibiotics was followed by a decline in the pace of antibiotics discovery in the 1990s. The key prerequisite for the resurgence of antibiotics research is the development of a computational discovery pipeline for antibiotics sequencing. We describe such pipeline for cyclic and branch cyclic peptides (cyclopeptides) that represent an important class of bioactive natural products such as antibiotics and anti-tumor compounds. Our CycloNovo algorithm for cyclopeptide sequencing reconstructed many new cyclopeptides that we validated with transcriptome, metagenome, and genome mining analyses. CycloNovo revealed a wealth of anti-microbial cyclopeptides from food that survive the complete human gastrointestinal tract, raising the question of how these cyclopeptides might affect the human microbiome.

bioinformatics↗

bioRxiv2018

American Gut: an Open Platform for Citizen-Science Microbiome Research

Although much work has linked the human microbiome to specific phenotypes and lifestyle variables, data from different projects have been challenging to integrate and the extent of microbial and molecular diversity in human stool remains unknown. Using standardized protocols from the Earth Microbiome Project and sample contributions from over 10,000 citizen-scientists, together with an open research network, we compare human microbiome specimens primarily from the USA, UK, and Australia to one another and to environmental samples. Our results show an unexpected range of beta-diversity in human stool microbiomes as compared to environmental samples, demonstrate the utility of procedures for removing the effects of overgrowth during room-temperature shipping for revealing phenotype correlations, uncover new molecules and kinds of molecular communities in the human stool metabolome, and examine emergent associations among the microbiome, metabolome, and the diversity of plants that are consumed (rather than relying on reductive categorical variables such as veganism, which have little or no explanatory power). We also demonstrate the utility of the living data resource and cross-cohort comparison to confirm existing associations between the microbiome and psychiatric illness, and to reveal the extent of microbiome change within one individual during surgery, providing a paradigm for open microbiome research and education.\n\nImportanceWe show that a citizen-science, self-selected cohort shipping samples through the mail at room temperature recaptures many known microbiome results from clinically collected cohorts and reveals new ones. Of particular interest is integrating n=1 study data with the population data, showing that the extent of microbiome change after events such as surgery can exceed differences between distinct environmental biomes, and the effect of diverse plants in the diet which we confirm with untargeted metabolomics on hundreds of samples.

microbiology↗