Hunting for microsatellite instability in long-read data with Owl

The paper introduces Owl, a Rust-based bioinformatic tool that leverages long-read sequencing data and a genome-wide panel of over 140,000 microsatellite markers to accurately quantify microsatellite instability, successfully identifying MSI-high cancers and revealing distinct motif-specific instability patterns, such as GGAA enrichment in Ewing sarcoma.

The Gist

Imagine your DNA is a massive, intricate instruction manual for building and running a human body. Now, imagine that manual has a few pages that are just lists of repeating words, like "ATATATAT" or "GGGGGG." These are called microsatellites.

In a healthy person, these repeating lists are very stable. But in some cancers, the "spell-checker" in the cell (called the Mismatch Repair system) breaks down. When this happens, the cell starts accidentally adding or deleting letters in those repeating lists. This chaos is called Microsatellite Instability (MSI).

Finding this chaos is crucial because if a cancer has high MSI, it often responds very well to powerful new drugs called immunotherapies.

The Problem: The "Blurry" Old Glasses

For years, scientists have tried to find this instability using short-read sequencing. Think of this like trying to read a long, repetitive sentence ("ATATATAT...") by looking at it through a pair of blurry, broken glasses. You can only see tiny fragments at a time. Because the glasses are blurry, it's hard to tell if a change in the text is a real mistake (a cancer mutation) or just a trick of the light (a sequencing error). Also, these old methods often require a "normal" sample from the patient to compare against, which isn't always available.

The Solution: Owl and the "High-Definition" Lens

This paper introduces a new tool called Owl.

1. The Owl's Eyes (Long-Read Sequencing)
Instead of blurry fragments, Owl uses long-read sequencing (specifically PacBio HiFi technology). Imagine this as switching from those blurry glasses to a pair of high-definition, 8K binoculars. Now, you can see the entire repeating sentence in one go. You can clearly see exactly how many "ATs" are there, without any guesswork.

2. The "Haplotype" Superpower
One of Owl's coolest tricks is that it can read the DNA while keeping the "mom" and "dad" versions of the chromosome separate (this is called phasing).

• The Analogy: Imagine a library with two identical copies of a book. If you mix the pages from both books together, you can't tell which page belongs to which book. But Owl keeps the pages sorted. This allows it to spot a mistake in one copy of the book without needing a second "normal" book to compare it to. This means Owl can detect cancer mutations even if the doctor only has the tumor sample, not the healthy tissue.

3. The "Wrap-Around" Algorithm
To count the repeats accurately, Owl uses a special math trick called "wrap-around alignment."

• The Analogy: Imagine you are counting the links in a giant chain. If the chain is too long to see all at once, you might lose count. Owl is like a chain-smith who can look at the chain, wrap their eyes around the loop, and count every single link perfectly, even if the chain is twisted or tangled.

What Did Owl Find?

The researchers tested Owl on 131 healthy people and 20 cancer samples. Here is what they discovered:

• It Works: Owl correctly identified the cancers with high instability (MSI-high) and matched the results of existing gold-standard tests.
• The "A" and "AT" Pattern: In most cancers, the instability happens mostly in short, simple repeats like "A" or "AT." Owl confirmed this.
• The Ewing Sarcoma Surprise: The most exciting discovery was in a specific type of bone cancer called Ewing Sarcoma.
  • The Analogy: While other cancers were messing up the "A" and "AT" words, these Ewing Sarcoma cells were specifically messing up the word "GGAA".
  • Why it matters: There is a known cancer-causing protein in these tumors that loves to grab onto "GGAA" sequences. Owl found that the instability wasn't random; it was happening exactly where this protein was hanging out, acting like a wrecking ball on the DNA instructions. This gives scientists a new clue about how this specific cancer works.

Why Should You Care?

Owl is like a new, super-smart detective for cancer.

1. It needs less evidence: It can find the crime (cancer) even if it only has the crime scene (tumor) and no witness (healthy tissue).
2. It sees the details: It doesn't just say "something is wrong"; it tells you exactly which repeating words are broken.
3. It helps choose the right medicine: By accurately spotting MSI, it helps doctors decide if a patient will benefit from life-saving immunotherapy.

In short, Owl takes the blurry, confusing picture of cancer genetics and turns it into a crystal-clear, high-definition map, helping doctors treat patients more effectively.

Technical Summary

1. Problem Statement

Microsatellite Instability (MSI) is a critical biomarker for mismatch repair (MMR) deficiency and a predictor of response to immunotherapy (e.g., checkpoint inhibitors). However, current detection methods face significant limitations:

• Short-Read Sequencing (SRS) Constraints: Existing tools (e.g., MSIsensor, DRAGEN) rely on short reads, which struggle to map complex or long repeats accurately. They often lack haplotype phasing, making it difficult to distinguish somatic mutations from germline heterozygosity without matched normal samples.
• Limited Marker Scope: Traditional assays (PCR or SRS-based) use small panels of specific markers (e.g., BAT-25/26), failing to capture genome-wide or motif-specific instability patterns.
• Missed Biological Signals: Standard assays focus heavily on homopolymers, potentially missing instability in other repeat motifs (e.g., GGAA) that may be biologically significant in specific cancer types like Ewing sarcoma.

2. Methodology: The Owl Tool

The authors present Owl, a bioinformatic tool implemented in Rust specifically designed to quantify MSI from PacBio HiFi long-read genomic data.

• Input Data: Requires aligned, haplotype-tagged BAM files (phased using tools like Hiphase or WhatsHap) and a BED file defining repeat regions and motifs.
• Workflow Modules:
  1. Profile Module:
     • Extracts reads spanning predefined microsatellite markers (1–6 bp motifs, 20–100 bp length).
     • Uses a wrap-around dynamic programming (WDP) algorithm to accurately count repeat copies within the read, handling the cyclic nature of repeats.
     • Groups reads by haplotype and calculates the Coefficient of Variation (CV) of repeat lengths for each locus. The CV is used to normalize variability relative to repeat length.
  2. Score Module:
     • Aggregates per-locus CV values across the genome.
     • Flags a marker as unstable if its CV exceeds a threshold (empirically determined as 5.0 based on a gamma distribution fit of control data).
     • Outputs the percentage of unstable markers (MSI score) and motif-specific breakdowns.
• Marker Set: Derived from GRCh38 RepeatMasker annotations, filtered for high sequence identity (>97%) and stability. The final set includes 146,562 high-quality markers across 131 diverse human genomes (HPRC) to establish baseline instability.

3. Key Contributions

• First Long-Read MSI Tool: Owl is the first tool optimized for long-read data, leveraging full-length repeat coverage and haplotype phasing to resolve somatic instability without requiring matched normal samples (tumor-only mode is viable).
• Genome-Wide & Motif-Specific Resolution: Unlike legacy panels, Owl analyzes >140,000 markers genome-wide, enabling the detection of instability patterns specific to different repeat motifs (homopolymers, dinucleotides, etc.).
• Discovery of Cancer-Specific Signatures: The tool revealed that while homopolymers and AT-rich dinucleotides are universal MSI markers, specific cancers (like Ewing sarcoma) exhibit unique instability patterns in GGAA motifs linked to the EWS::FLI1 fusion protein.
• Integration: Owl is integrated into the PacBio HiFi Somatic workflow, providing a scalable framework for cancer genomics.

4. Results

• Baseline Stability: In 131 healthy control genomes (HPRC), the background MSI score ranged from 1.4% to 5.4% (mean ~2.2%), establishing a robust baseline for distinguishing somatic instability.
• Cancer Detection: Applied to 19 cancer cell lines and one tumor-normal pair, Owl identified 5 MSI-high samples (Ewing sarcoma TC32/CHLA10, diffuse astrocytoma, and two gastric cancers) with instability scores of 15–18%.
• Validation: In a diffuse astrocytoma sample, Owl's results (16.67% unstable sites) showed close concordance with the Illumina DRAGEN MSI assay (20% unstable sites), validating long-read accuracy.
• Motif-Specific Insights:
  • Universal Signal: Homopolymers (A/T) and AT-rich dinucleotides showed the highest instability across all MSI-high samples.
  • Ewing Sarcoma Signature: Two Ewing sarcoma lines (TC32, CHLA10) displayed a distinct pattern of elevated instability in GGAA motifs (and close edit-distance variants). These motifs were significantly enriched in enhancer elements and genes regulated by EWS::FLI1 (e.g., EZH2, SOX5, SOX6). This suggests a mechanistic link between the fusion protein's binding and local microsatellite instability, a signal missed by standard homopolymer-focused assays.

5. Significance

• Clinical Utility: Owl provides a more accurate, phasing-aware method for MSI detection that can operate on tumor-only samples, addressing a major logistical hurdle in clinical settings where matched normals are unavailable.
• Biological Discovery: By moving beyond short-read limitations, Owl uncovers novel biological mechanisms, such as the specific role of GGAA repeat instability in Ewing sarcoma pathogenesis.
• Future Directions: The framework sets the stage for exploring RNA-level instability signatures and expanding to targeted long-read panels for higher throughput. It represents a shift from limited marker panels to comprehensive, genome-wide instability profiling in cancer.