SQANTI-browser: visualization and curation of SQANTI3-classified long-read transcriptomes within the UCSC Genome Browser

SQANTI-browser is a novel visualization framework that integrates SQANTI3-classified long-read transcriptome data into the UCSC Genome Browser, enabling interactive filtering, evidence-guided curation, and the resolution of alignment artifacts to rescue actionable novel isoforms across diverse datasets.

The Gist

Imagine your DNA is a massive, ancient library containing the instructions for building a human. For a long time, scientists could only read these instructions in short, choppy snippets (like reading a book one word at a time). But now, we have a new technology called long-read sequencing that lets us read entire chapters or even whole books in one go. This is amazing because it helps us discover new, complex versions of stories (called "isoforms") that we never knew existed.

However, there's a catch: because these new "books" are so long and complex, they are often messy. They might have typos, missing pages, or parts that look like they belong to a different story entirely. It's like trying to organize a pile of shredded manuscripts where some pieces are from the same book, some are from different books, and some are just random scraps of paper. This is the "technical and structural ambiguity" the paper talks about.

Enter SQANTI-browser.

Think of SQANTI3 as a very strict librarian who has already sorted through this messy pile and stamped every manuscript with a label saying, "This is a real story," "This is a fake," or "This is a weird mix-up." But until now, looking at those labels was like reading a boring spreadsheet.

SQANTI-browser is the tool that takes those labels and turns them into a high-tech, interactive map that you can explore inside the famous UCSC Genome Browser (which is like Google Maps for DNA).

Here is how it works in simple terms:

• The Visual Guide: Instead of just looking at a list of data, you can now "fly over" your DNA map. You can see exactly where a story starts and ends, and you can instantly see the librarian's stamp (the classification) right next to it.
• The Filter: Imagine you are looking at a crowded city map. SQANTI-browser lets you put on special glasses that hide all the "fake" stories and only show you the "real" ones, or vice versa. This helps scientists quickly find the important new stories without getting lost in the noise.
• The Detective Work: It allows scientists to compare these new stories against public libraries (other known data) to see if they match up or if they are truly new discoveries.
• The Flexible Toolkit: The paper notes that this tool is very adaptable. It doesn't just work for standard human DNA; it can handle "non-reference" genomes (like unique or mutated versions) and can be customized with extra notes, just like adding custom stickers to a map.

What did they prove?
The authors tested this tool on three specific types of messy data:

1. Clinical data: Real-world samples that might be a bit "noisy" or difficult to read.
2. Noisy datasets: Data that is intentionally difficult to simulate real-world problems.
3. Synthetic datasets: Made-up data created specifically to test the tool's limits.

In all these cases, SQANTI-browser acted like a skilled editor. It helped scientists spot and remove "alignment artifacts" (which are like optical illusions or mirages in the data that look like real stories but aren't). More importantly, it helped them "rescue" actionable novel isoforms—meaning it saved real, useful new stories that might have been thrown away as mistakes by other methods.

In short: SQANTI-browser turns a confusing, messy pile of long-read DNA data into a clear, interactive, and filterable map, helping scientists distinguish between real biological discoveries and digital noise.

Technical Summary

Technical Summary: SQANTI-browser

The Problem
While long-read sequencing has revolutionized transcriptome-wide isoform discovery, it introduces substantial technical and structural ambiguity that complicates the accurate interpretation of transcripts. Although tools like SQANTI3 exist to classify these transcripts, the resulting data often lacks an intuitive, interactive environment for researchers to visually inspect, filter, and curate complex structural metadata alongside standard genomic annotations. This gap hinders the effective resolution of alignment artifacts and the identification of biologically relevant novel isoforms.

Methodology
To address these challenges, the authors present SQANTI-browser, a classification-aware visualization framework designed to bridge SQANTI3 outputs with the UCSC Genome Browser ecosystem. The core methodology involves:

• Data Conversion: Transforming SQANTI3 classification outputs into interactive UCSC Genome Browser Track Hubs.
• Metadata Preservation: Ensuring that full transcript structural metadata is retained and accessible within the visualization.
• Integration: Embedding SQANTI classifications directly into the UCSC interface, allowing users to view these classifications alongside public resource tracks.
• Adaptive Architecture: Implementing a flexible system that natively supports non-reference genomes, orthogonal data types, and custom metadata fields.

Key Contributions
The primary contribution of this work is the creation of a dynamic environment that enables evidence-guided curation. By integrating classification data directly into a familiar genomic browser, SQANTI-browser allows for:

• Dynamic Filtering: Users can filter transcripts based on SQANTI3 classification categories in real-time.
• Contextual Curation: Researchers can curate transcripts while simultaneously viewing supporting evidence and public annotations.
• Artifact Resolution: The framework facilitates the visual identification and resolution of alignment artifacts that might otherwise be misinterpreted as novel biology.

Results
The authors applied SQANTI-browser to a diverse set of datasets, including:

• Clinical datasets
• Noisy datasets
• Synthetic datasets

In these applications, the tool successfully resolved alignment artifacts and "rescued" actionable novel isoforms that might have been overlooked or discarded in standard analysis pipelines. The results demonstrate the tool's capability to handle the complexity and noise inherent in long-read transcriptome data.

Significance
The paper positions SQANTI-browser as a robust framework for long-read transcriptome curation. Its significance lies in its ability to convert complex, classification-based data into an interactive, visual format that supports rigorous manual review. By preserving structural metadata and enabling dynamic filtering within the UCSC ecosystem, the tool provides a practical solution for interpreting long-read data, ultimately aiding researchers in distinguishing technical artifacts from biologically meaningful isoforms.