CNV Stability Without Selection Pressure

Code and data associated with De et al. 2025. Each figure script contains all required data loading, statistical analysis, and visualization for the corresponding figure in the paper.

Preprint: https://doi.org/10.1101/2025.07.21.665951

---

Repository Structure

.
├── Fig1.py                          # Figure 1: CNV strain schematics
├── Fig2.py                          # Figure 2: Relative fitness of CNV strains
├── Fig3.py                          # Figure 3: CNV dynamics across LTEE lineages
├── Fig4.py                          # Figure 4: Phases of CNV loss and time-to-loss
├── Fig6.py                          # Figure 6: Fitness of revertants vs. CNV parents
├── Fig7.py                          # Figure 7: PacBio read depth and flow cytometry
└── Data/
    ├── Fig_Data/                    # Input data files for all figures
    │   ├── *.csv                    # Tabular data (see below)
    │   ├── *.fcs                    # Flow cytometry files (Figure 7)
    │   └── Bedgraph_files           # Link to externally hosted bedgraph/SAM files
    └── Read_Depth_Plots_Revertants/ # Read-depth plots for revertant strains (PNG)

---

Figure Scripts

Fig1.py — CNV Strain Schematics

Produces two panels:

• Fig 1A: Schematic illustration of the three possible outcomes for CNV frequency over time — selection for CNVs, stable CNVs, or loss of CNVs — drawn as sigmoid/flat/sigmoidal curves with added noise.
• Fig 1B: Genomic maps of all CNV strains used in the study, drawn as colored filled regions on chromosomes XI (GAP1 locus), XIV (MEP2 locus), and XV (PUT4 locus). A fourth panel shows double-aneuploid strains carrying CNVs on both Chr XI and Chr XIV.

Key inputs: All_strain_schematic_GAP1_20250122.CSV, All_strain_schematic_MEP2.CSV, All_strain_schematic_PUT4.CSV

---

Fig2.py — Relative Fitness of CNV Strains

Produces three panels:

• Fig 2A: Raw CNV frequency (percentage of cells with CNV) across 15 generations of pairwise competition for three replicates of strain GM_13 (DGY2750).
• Fig 2B: Log-ratio of strain to ancestor frequency over time, with a linear regression fit and 95% confidence interval. The slope estimates relative fitness.
• Fig 2C: Scatter plot of CNV fitness cost (calculated from gene expression data) vs. measured relative fitness across all CNV strains, with a Monte Carlo regression and 95% CI band. Tests whether larger/costlier CNVs are less fit.
• Fig 2D: Bar chart of relative fitness (slope + 1) for all CNV strains, with error bars (2× SE) and significance stars for strains with p ≤ 0.05.

Statistical methods: Linear regression (scipy.stats.linregress), Monte Carlo simulation (10,000 iterations) for regression uncertainty, Pearson correlation.

Key inputs: Pairwise_III_edited_new.CSV, GaschFitnessCost_vs_Size.CSV, Pairwise_Regression_Norm_All_30C_final.CSV

---

Fig3.py — CNV Dynamics Across LTEE Lineages

Produces three multi-panel figures tracking the percentage of cells carrying a CNV over hundreds of generations in a long-term evolution experiment (LTEE):

• Fig 3A: Four-panel plot showing CNV dynamics for three GAP1-CNV strains (G_1, G_2, G_3) and control strains (multi-copy and one-copy). Each main panel shows LTEE data; each inset shows the ancestral evolution experiment from which the strain was isolated.
• Fig 3B: Eight-panel plot (2 × 4 grid) showing dynamics for eight additional strains: four GAP1-CNV strains (G_4–G_9), one MEP2-CNV strain (M_10), and one PUT4-CNV strain (P_11).
• Fig 3C: Eight-panel plot (2 × 4 grid) for additional MEP2 and PUT4 CNV strains plus double aneuploids (strains GM_12–GM_15).

Key inputs: 2024_11_Percentage_gating_LTEE3_YPD_20241205_edited.csv, 2023_12_Percentage_gating_LTEE2_edit.csv, 2018_Percentage_CNV_acquire_Lauer.csv, 2018_Percentage_CNV_acquire_Rahman.csv, 2018_Percentage_CNV_acquire_Avecilla.csv

---

Fig4.py — Phases of CNV Loss and Time-to-Loss

Produces a four-panel figure examining when CNV loss occurs during the LTEE:

• Fig 4A: Example trajectory (DGY1886_1) with a cubic spline fit, annotated with the 75%, 50%, and 25% CNV-frequency thresholds that define the early, middle, and late phases of CNV loss.
• Figs 4B–4D: Box plots comparing the generation at which CNV frequency drops below 75% (early), 50% (middle), and 25% (late) for three CNV classes — Chr XI segmental (G_5), Chr XI aneuploid (GM strains), and Chr XIV aneuploid (GM strains). Points exceeding 220 generations are plotted as triangles above the y-axis; sample sizes are annotated. Significance brackets show Welch's t-test p-values between groups.

Statistical methods: Cubic spline interpolation (scipy.interpolate.CubicSpline), root-finding (scipy.optimize.root_scalar), Welch's t-test (scipy.stats.ttest_ind), pairwise comparisons across all three CNV classes.

Key inputs: 2023_12_Percentage_gating_LTEE2_edit.csv

---

Fig6.py — Fitness of Revertants vs. CNV Parents

Produces a single bar chart comparing relative fitness (slope + 1 from pairwise competition assays) for all CNV parent strains and their revertants (both complete and partial):

• Bars are colored by type: grey for CNV strains, light grey for complete revertants, hatched for partial revertants.
• Error bars show 2× SE; dotted vertical lines separate strain groups (G_4, GM_12–GM_15).
• Cyan shaded rectangles on CNV-parent bars show fitness ranges estimated by simulation-based inference (SBI).

Key inputs: Pairwise_Regression_Norm_All_30C_final.CSV

---

Fig7.py — PacBio Read Depth and Flow Cytometry

Produces a figure combining genomic and phenotypic data from a CNV parent strain (DGY1886) and its revertants (DGY3417, DGY3410, DGY3414, DGY3418, DGY3420):

• Left panels (ax1–ax3): PacBio read depth across Chr XI (positions 400–650 kb), normalized to ploidy, with a Gaussian smoothing filter (σ = 20). Shows the CNV parent (ax1), partial revertant (ax2), and complete revertants (ax3, four strains overlaid). Vertical dotted lines mark CNV breakpoints; split reads (from SAM files) indicating structural variant junctions are plotted as red tick marks on ax1 and ax2.
• Right panel (ax4): Flow cytometry scatter plot (FSC-A vs. log mCitrine fluorescence) for all six strains, used to confirm CNV copy number by fluorescence intensity. Data are loaded from .fcs files using FlowCal.

External data: Bedgraph and SAM files are large and hosted externally; see Data/Fig_Data/Bedgraph_files for the download link.

Key inputs: PacBio_*.bedgraph and Split_reads_*.sam (external), DGY*.fcs

---

Data Files

All tabular data files are in Data/Fig_Data/:

File	Description	Used in
All_strain_schematic_GAP1_20250122.CSV	Genomic coordinates (start/end) and copy number for all GAP1-CNV strains on Chr XI	Fig 1
All_strain_schematic_MEP2.CSV	Genomic coordinates for MEP2-CNV strains on Chr XIV	Fig 1
All_strain_schematic_PUT4.CSV	Genomic coordinates for PUT4-CNV strains on Chr XV	Fig 1
Pairwise_III_edited_new.CSV	Pairwise competition data (% CNV by generation) for strain DGY2750 replicates	Fig 2
GaschFitnessCost_vs_Size.CSV	Calculated CNV fitness costs and measured fitness slopes/SE for all strains	Fig 2
Pairwise_Regression_Norm_All_30C_final.CSV	Linear regression results (slope, intercept, SE, p-value, type) for all strains and revertants	Figs 2, 6
2024_11_Percentage_gating_LTEE3_YPD_20241205_edited.csv	Flow cytometry gating data (% multicopy mCitrine/mCherry) for LTEE3 strains by generation	Fig 3
2023_12_Percentage_gating_LTEE2_edit.csv	Flow cytometry gating data for LTEE2 strains by generation	Figs 3, 4
2018_Percentage_CNV_acquire_Lauer.csv	Historical CNV acquisition data (Lauer et al. 2018) used for inset plots	Fig 3
2018_Percentage_CNV_acquire_Rahman.csv	Historical CNV acquisition data (Rahman et al. 2018) used for inset plots	Fig 3
2018_Percentage_CNV_acquire_Avecilla.csv	Historical CNV acquisition data (Avecilla et al. 2018) used for inset plots	Fig 3
Bedgraph_files	Text file containing a Google Drive link to PacBio bedgraph and split-read SAM files	Fig 7
DGY*.fcs	Flow cytometry FCS files for CNV parent and revertant strains	Fig 7

Data/Read_Depth_Plots_Revertants/ contains read-depth PNG plots for individual revertant strains on Chr XI and Chr XIV (named Chr11_<DGY#>.png and Chr14_<DGY#>.png).

---

Dependencies

numpy
matplotlib
pandas
scipy
statistics (standard library)
FlowCal        # Fig7 only — flow cytometry FCS file parsing

Install with:

pip install numpy matplotlib pandas scipy FlowCal

pysam is imported in Fig7.py but not required (SAM files are parsed directly with pandas).

---

Usage

Each script is self-contained. Run from the directory containing the data files, or update the file paths at the top of each script. Output is displayed interactively; commented-out plt.savefig() calls at the bottom of each script can be uncommented to save figures to disk.

cd Data/Fig_Data
python ../../Fig1.py