henhaobio
Sign in

bioRxivNeurosciencePreprint

A novel analysis pipeline and experimental design for measuring expanded repeat instability

Christopher SmithiD, Ruban Rex Peter DurairajiD, Emma L. RandalliD, Alys N AstoniD, Laura HeratyiD, Waheba ElsayediD, Alvaro MurilloiD, Vincent DioniD

PDFDOICodeData
In-depth full-text digest available · Jump to digest ↓

Abstract

The expansion of short tandem repeats is a feature of over 60 different human diseases. Ongoing somatic instability throughout a patients lifetime can influence disease progression and has emerged as a therapeutic target. Understanding its mechanism is essential for the identification of both drug targets and therapeutic interventions. A major obstacle towards this translational goal has been to measure changes in repeat size distributions. To address this we provide a new analysis method and accompanying software that generates delta plots, extracts the instability frequency from targeted long-read sequencing data, the instability bias, and the average size of the changes. It further provides statistical analysis for comparison between treatments. We show its applicability to non-dividing cells and in vivo datasets. Moreover, we have developed a streamlined experimental design for dividing cells, Single Clone-based Instability Assay (SCIA), that saves weeks in assessing the effect of a gene knockout on repeat instability and is ideal for an initial screen. We have validated the approach using FAN1, PMS1, and MLH1 knockouts in cell lines and induced pluripotent stem cells. We find that FAN1 knockout clones showed, as expected, an increased frequency of repeat expansion but the size of the expansions were smaller. This highlights the wealth of information that can be extracted and the potential for novel insights into the mechanism of repeat instability.

In-depth summary

Claude Opus 5.5 · xhigh effort · 7 Nov 2026 · from the full text of version 3 (8 Oct 2026) · CC BY 4.0

Delta plots from targeted long-read data separate repeat instability frequency, expansion bias and change size; a single-clone assay (SCIA) shows FAN1 knockout raises expansion frequency but shrinks expansions.

The widely used instability index merges the size and frequency of repeat changes into one number, which the authors say hides mechanism and lowers statistical power. This work offers delta plots with a GUI and statistics, plus a single-clone design that avoids long bulk cultures and reporter assays.

Key results

  1. ShownDelta plots recover the expected accumulation of expansions over time in Huntington's disease (HD) iPSCs, and the pipeline also ran on iPSC-derived neurons and mouse brain and tail datasets.

    HD iPSCs: instability frequency 10.0% at day 21 and 18% at day 42; bias ratio 15.5 then 57.1 expansions per contraction; mean expansion size at day 21 was 6.2 ± 9 repeats. · Quantification of repeat size mosaicism in…

  2. ShownSCIA on GFP(CAG)116 cells gives a multi-parameter view of instability. In one bulk culture, a short allele took over within a month, an artefact not seen in single-cell-derived cultures.

    Over 42 days: 12.8% of alleles changed, bias ratio 4.25, mean expansion 25.7 ± 11 CAGs, mean contraction -39.6 ± 24. Up to 12 clones per assay were judged manageable. · Minimising the effect of clonal expansion in…; Single Clone-based Instability Assay (SCIA)

  3. ShownSCIA separates repeat length effects on frequency from effects on size. The 116-repeat line had a higher expansion bias and larger changes, but a slightly lower instability frequency.

    Two-tailed KS test P=0.0005 between lines; bias ratio 1.48 (81 repeats) vs 4.25 (116); frequency 16.9% (81) vs 12.8% (116); change size nearly four times larger in the longer line. · Reproducing the effect of repeat size on repeat…

  4. ShownFAN1 knockout clones had more frequent but smaller expansions than the parent line.

    Two independent FAN1 knockout clones differed from the parent (P ≤ 0.0087): frequency 29.2% and 36.1% vs 12.8%; mean expansion 16.3 and 14.1 CAGs vs 25.7. · Validation of SCIA using genetic knockout lines

  5. ShownPMS1 and MLH1 knockouts shifted the bias towards contractions with little change in instability frequency; MLH1 was tested with a faster protocol that skips pre-made clones.

    PMS1 knockouts (two clones, P<0.0001): bias ratio fell from 4.25 to below 1, frequency unchanged. MLH1 knockout (P<0.0001): frequency 27.0% vs 22.3% for parent, bias ratio 0.08. · Validation of SCIA using genetic knockout lines; A quick protocol to evaluate the effect of a…; Discussion

Methods, data and reproducibility

Repeat sizes are read from targeted PacBio long-read data with Repeat Detector; normalised histograms are aligned to the day 0 mode and subtracted to give delta plots, from which frequency, bias ratio and mean change size are computed. SCIA grows single-cell clones for 42 days and compares each with the day 0 distribution.

  • HD iPSC time course (days 0, 21, 42), targeted long-read sequencing of the HTT repeat · human · PRJNA1428164
  • HD iPSC-derived neurons and R6/1 mouse striatum with Cas9D10A nickase, re-used from prior work · human; mouse · PRJNA1077893
  • GFP(CAG)81 and GFP(CAG)116 HEK293-derived reporter cells, SCIA clones · human (HEK293-derived) · n = up to 12 clones per assay; 8 to 12 colonies targeted · PRJNA1428164
  • FAN1, PMS1 and MLH1 knockout lines in GFP(CAG)116 background · human (HEK293-derived) · n = 2 clones each for FAN1, PMS1; MLH1: 10 knockouts of 21 clones screened · PRJNA1428164
Methods & tools
Repeat Detector, PacBio SMRT sequencing (Sequel IIe), SeqKit, wxPython, pandas, NumPy, seaborn, matplotlib, Docker, Kolmogorov-Smirnov test, Mann-Whitney U test, ImageJ
Software & environment
Software v2.0: Python 3.13.5, wxPython 4.2.3, pandas 2.2.3, NumPy 2.1.3, seaborn 0.13.2, matplotlib 3.10.8, Docker 28.4.0; Singularity object for HPC; tested on macOS.
Reproducibility gaps
Requires Docker and a separate repeat-detector.tar file. Some scripts were written with AI help. Primers, knockout designs and antibodies are in supplementary tables not shown here. New data are cited only as SRA PRJNA1428164, without a URL.
Limitations (5)
  • The software cannot average replicates that each have their own control and treated samples; this must be done by hand. · stated by the authors
  • The pipeline needs a treatment or time that changes repeat size; a single sample from one time point or individual gains nothing from it. · stated by the authors
  • The quick protocol depends on highly efficient Cas9 RNPs: FAN1 yielded only 2 homozygous knockouts from 53 clones, so the shortened timeline could not be used. · stated by the authors
  • It is unclear whether the finding of similar instability frequency at 81 and 116 repeats extends to other repeat sizes or cell types. · stated by the authors
  • Validation uses one HEK293-derived reporter family with two knockout clones per gene (MLH1 from a bulk population). There is no simulation or ground-truth benchmark of the delta-plot metrics, only agreement with known biology. · noted in this summary

Read this if you study short tandem repeat instability and want delta-plot metrics from long-read data, or a faster single-clone knockout screen in cultured cells.

Citations
1

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISKim '26Dolzhenko '26Sanden '24Taylor '22McGinty '25Hujoel '26
59% semantic matchBioinformatics
Junsoo Kim·bioRxiv·Mar 18, 2026