Skip to main content
Premium Trial:

Request an Annual Quote

One Biosciences Using Single-Cell Transcriptomics to Understand Cancer Therapy Response, Resistance


NEW YORK – One Biosciences aims to refine single-cell analysis for diagnostic and therapeutic purposes through a novel method of multimodal pathway analysis.

The Paris-based company spun out from the Institut Curie in 2020 with the plan to discover biomarkers and new therapies through single-cell analysis. Helping to drive this analysis is an algorithm called Maya, which was recently published in Nature Communications.

Maya looks for features that disease-associated cells have in common across different patients and tissues via a unique pathway analysis method that detects how each pathway is activated across cell types, where each "mode" of activation relies on distinct subsets of genes. The algorithm provides a score describing the activity of each mode and groups genes according to their similarity in those scores.

In the Nature paper, for instance, Maya identified several combinations of genes characterizing the epithelial-to-mesenchymal transition (EMT), which often enables solid tumors to become more invasive and therefore metastatic, in various cell types. The gene groups underlying EMT varied by cell type, with macrophages, cancer-associated fibroblasts, and epithelial cells all using different gene sets to achieve the same outcome.

This method contrasts with prior tools such as AUCell and pagoda2, which are "very powerful" but give "only one score per cell for the entire dataset as if every cell activates [a pathway] the same way," said Céline Vallot, a One Biosciences founder and a group leader at the Institut Curie.

Analyzing single-cell RNA-seq data by pathway also provides a way to boost the signal-to-noise ratio per cell by essentially deriving multiple data points from each gene, since these all participate in multiple pathways.

"Maya is a very interesting pathway activity scoring algorithm," Jiliang Tang and Yuying Xie, computer scientists at Michigan State University whose research focuses on single-cell data analysis, said in an email.

Its ability to detect multiple modes per pathway, they added, marks an advance in the field of single-cell and pathway analyses.

"It is very important to have the ability to define multiple scores per pathway, because there are multiple ways to define a pathway," they wrote. "Some are based on [gene ontology] gene sets, some are based on differentially expressed genes, some are based on protein interactions, etc. As a result, different sets of genes within a pathway may have different biological functions in different cell types or under different conditions. Using Maya to decouple them provides more flexibility and granularity to pathway analyses."

Tang and Xie also mentioned that although the paper describing Maya did not address spatially resolved single-cell data, they could "clearly see its potential" in analyzing those datasets, which tend to be noisier and even more sparse than single-cell RNA-seq data.

Grouping genes in biologically related pathways and analyzing them together, they added, can increase the signal-to-noise ratio in typically very noisy single-cell datasets to provide more robust results.

These results, Vallot explained, ultimately enable researchers to peer through patient-specific effects that mask shared disease-specific biological features.

Patient-specific effects seen in single-cell datasets, Vallot said, are often driven by things such as copy number variation or other DNA alterations that make the data look different even though the underlying disease biology may be the same.

"All the genetics hides the complex biology, [such as] pathways," she said.

Vallot noted that while Maya appears quite effective in identifying commonalities in patient-specific data, batch effects that affect all cell types in a sample can present a challenge.

For Maya to work, she said, "there needs to be some variation in the data."

Since being spun out in 2020, One Biosciences has focused on developing its single-cell data mining and in-depth characterization protocols in anticipation of working with hospitals and pharmaceutical companies to discover novel biomarkers of and therapies for cancer and other diseases.

"The idea is to partner up with a hospital and to make a deal to have access to samples [with which] to generate large-scale single-cell data," Vallot said, adding that "we're most interested in response and resistance to therapy."

One Biosciences has currently partnered with the Institut Curie Hospital to study response and resistance in ovarian cancer, and with the Hartmann Clinic, Paris Saint-Joseph Hospital, and the Rafael Institute Research Center, all Paris-based institutions, to do the same with laryngeal cancer, although the company does not yet have a lead molecule in its pipeline.

Vallot said that the company also has "quite a lot" of ongoing pharmaceutical collaborations, although the pharma partners remain confidential for the time being.

The company licensed its core technologies, which mainly consist of unnamed methods for conducting single-cell transcriptomics on frozen patient samples, from the Institut Curie, the Sorbonne, and France's National Center for Scientific Research (CNRS).

Vallot cofounded the company using technologies developed in her own and other labs at Curie, where she remains as a group leader while serving as scientific adviser to One Biosciences.

"We continue to do some co-innovation with the company," she said.

The company, which currently employs "10 or 11" people, plans to initiate fundraising in mid-2024.