The Weaver Plot: Analyze all gene data from multiple studies in a single view
Subscribe to stay up to date with the latest Sonrai content.
Author: Matthew Chung, Bioinformatician
To achieve breakthroughs in cancer treatment. Understanding the biology of biomarkers and therapeutic targets is critical. Researchers often face the challenge of interpreting data from both internally generated and externally sourced sources, with each dataset having its own unique scale and methodology.
For example, analyzing a tissue sample can involve different techniques, such as RNA sequencing to profile transcript expression or mass spectrometry to measure protein levels. Differences in sample processing and data handling across studies add to the complexity, leaving scientists and clinicians with scattered data, like pieces of a puzzle spread across different tables.
There’s no easy way to compare, interpret and view these varied data sets side by side to form a comprehensive picture and draw meaningful conclusions from the data. The Weaver plot is our latest tool designed to address this gap by providing a unified data comparison and analysis platform.
Imagine What You Could Discover
Our cloud-based AI technology unifies and integrates all data sources from discovery, research, and development to regulatory approval.See It Now
The Weaver Plot:
The Weaver plot, aptly named for its ability to interlace data threads into a comprehensive tapestry of insights, addresses these challenges. It allows for the integration of multimodal data like proteomics and RNA-seq into a coherent visualization. This tool is invaluable for generating insights, identifying knowledge gaps, and formulating hypotheses.
Case Study: Whole Body Gene Expression:
Our showcase focuses on understanding the normal distribution of biomarkers and therapeutic targets by examining whole-body gene expression in healthy states. The Weaver plot aids in overcoming inconsistencies in tissue naming and classification across studies, using an ontology mapping algorithm and an anatomy ontology with over 1,400 human anatomical entities. For example, one study may have used the term ‘brain’, while another chose ‘cerebrum’. By mapping tissues to the ontology, we ensured consistency and addressed varying anatomical granularity, which maximized the number of comparable tissues across datasets.
We also developed a ranking algorithm that derives a consensus rank score to allow users to quickly generate insight from multimodal studies. Thus, with the Weaver plot, diverse datasets of healthy tissues can be compared and analyzed in unison despite their different origins and methods. This harmonization is key to identifying the most promising targets for cancer therapy.
How the Weaver Plot Works:
In our demonstration, we focused on 370 prioritized gene targets for cancer identified through CRISPR screens and a clinically informed cancer dependency map. We gathered seven publicly available proteomics and RNA-seq datasets, harmonizing tissue classification via our ontology algorithm, and selected 41 relevant tissues for this analysis.
To use the Weaver plot, users can select the gene and the algorithm will automatically extract and harmonize the corresponding protein or RNA expression data irrespective of original units or methods.
For example, we can choose ALK from the category of targets that has ‘approved or clinical preclinical drugs’, a receptor tyrosine kinase associated with non-small cell lung cancer, and anaplastic large cell lymphomas to reveal its expression data across multiple tissues.
The plot structure includes a bar plot for each data set, each bar representing a tissue type with the leftmost bar plot showing the aggregated rank score.
Slope graphs connect tissue rankings across datasets, enabling a detailed understanding of factors contributing to the final tissue ranking. In the case of ALK, under the default setting, the brain tops the ranking. Clicking on a bar or line highlights all corresponding data for that tissue.
Hovering over a bar reveals the exact ALK expression value in that tissue. The Weaver plot doesn’t just show data, but also shows its depth, highlighting tissues with limited data sources.
For example, the pituitary gland, second in ALK expression, is marked in red. If we click on it, we can quickly see that there is only one RNA-seq evidence available.
The Weaver plot also distinguishes between zero expressions and non-assessed tissues for accurate interpretation.
For example, inspecting column data reviews high expression in some datasets but non-detection in others. This guides the user to do a deep dive to examine the evidence.
User Control and Insight Generation:
Users have control over the analysis process, including the ability to exclude certain datasets or adjust data type weightings.
For example, if the user would like to exclude these two datasets from this analysis, we simply uncheck the datasets from the list, then proceed to update the plot.
Now the ranking of COLON has become much higher. In some cases, the user may want to emphasize proteomics over transcriptomics data.
We can lower the weighting for RNA data. Now COLON has the highest rank after we gave proteomics evidence more importance.
Based on the available evidence, if we are developing a drug targeting ALK in non-small cell lung cancer, we may want to look out for potential toxicity issues, particularly in the brain and colon.
While indirect, it is certainly a critical insight to guide experimental strategies and research directions.
Conclusion and Invitation:
The Weaver plot is not just a tool; it’s a gateway to deeper insights and more effective cancer research strategies. If this aligns with your research needs, we invite you to explore its potential and enhance your research endeavors. Reach out to us to discover how the Weaver plot can revolutionize your approach to cancer research.
Get in touch
Like What You See? Let’s Talk
No hard sales conversations We Listen to your problems We give you confidence to make your decision
Take The First Step