This repository contains analysis code for the project 'Overnight circuit remodelling enables the onset of caregiving in juveniles' by Bradley B. Jamieson, Maxwell X. Chen, Lina S. H. El Rasheed, Phoebe M. Claxton, Swang Liang, Patty Wai, Grace M. K. Chattey, Maria Voronkov, Emma Davis, J. Mark Skehel, I. Lorena Arancibia-Cárcamo, Molly Strom and Johannes Kohl

ANOVA_vsAll Loads a CSV of group columns (e.g., P14, P15), reshapes to long format, and checks normality across groups (Shapiro–Wilk, ≥3 values). For every pair of groups it runs either parametric ANOVA + pairwise t-tests (Holm-corrected) or Kruskal–Wallis + Dunn post-hoc (Holm-corrected), then exports a *_results.csv with tests, p-values, and significance stars.

ANOVA_vsControl Compares a user-specified control column against every other group column after reshaping to long format and assessing normality. Uses OLS ANOVA with a control-vs-group multiple-comparisons adjustment when parametric, or Kruskal–Wallis + Dunn (Holm) when nonparametric, and saves a *_results.csv summary with corrected p-values and stars.

BrainGlobe_Processing Reads a BrainGlobe/Cellfinder summary table of per-structure cell counts and volumes and removes a predefined list of excluded structures (e.g., fibre tracts/ventricles/broad parents). Computes cell density (cells per mm³), aggregates by structure, and exports a cleaned summary to Excel.

Chi2_Independence Loads a contingency table CSV, coerces to numeric, drops missing rows, and runs a chi-square test of independence between two columns (groups). Reports raw and BH-FDR–adjusted p-values (plus stars) and saves results to *_chi2_results.csv.

Discrimination Takes a stimulus-by-response CSV (columns = stimuli), computes one-vs-rest ROC AUC per stimulus with bootstrap mean ± SEM, and builds a pairwise dominance matrix P(AUC_A > AUC_B). It also computes a bootstrapped tuning specificity index per stimulus, tests TSI > 0 with Wilcoxon, applies BH-FDR correction, and exports all summaries to a single results CSV.

Elbow_points Loads and concatenates morphometric CSVs, keeps user-selected features, standardizes them, and fits K-means across K=1–10. Uses the curvature (“elbow”) of inertia to pick an optimal K, plots the elbow curve, and prints the chosen K.

EphysAnalysis Reads electrophysiology ABF files and associated metadata to analyse resting membrane potential, spontaneous firing, IVCC responses, and rheobase ramps. Quantifies spike frequency, firing patterns (silent, sparse, tonic, irregular), action potential properties, input resistance, membrane time constant, sag, rebound, and adaptation. Exports per-cell summaries, individual spike measurements, and firing classifications to Excel.

Fisher's_multiple Reads a 2×N In/Out count table and runs Fisher’s exact test on adjacent column pairs (1 vs 2, 3 vs 4, etc.). Applies BH-FDR across all tests, assigns significance stars, and saves a results CSV with raw and corrected p-values.

Fisher's_vsControl Loads an In/Out count table and compares each group column to a specified control column using Fisher’s exact test. Outputs raw p-values, applies BH-FDR correction (as implemented in the script), assigns stars, and saves a *_results.csv summary.

Gene_ANOVA Parses a gene table where columns encode Region and Age (Region_Age), reshapes to long format, and fits a two-way ANOVA (Region * Age). It then performs within-region pairwise age comparisons, applies BH-FDR across all post-hoc tests, and exports ANOVA + post-hoc results to CSV.

KaplanMeier_multiple Loads retrieval-time columns, reshapes to long format, and defines an event as retrieval < 600 (≥600 censored). Runs log-rank tests for adjacent column pairs, applies BH-FDR across comparisons, and saves a *_results.csv with corrected p-values and stars.

KaplanMeier_vsControl Reshapes retrieval-time data and treats Vir. as the control group, with events defined as retrieval < 600. Runs log-rank tests comparing Vir. to each timepoint, BH-FDR–corrects across tests, and exports a *_results.csv summary.

Linear_regression Fits an OLS linear regression between two numeric columns from a CSV and reports slope, intercept, R², and slope p-value. Generates a scatter plot with fitted line and an approximate 90% confidence band, and saves a small results CSV.

MassSpec_CellChatDB (R) Filters a hypothalamus DE protein table (FDR < 0.05, |log2FC| > 1) and uses it to subset CellChat’s mouse ligand–receptor interaction database. It also plots selected genes across region × age groups using expression matrices plus metadata, saving gene-wise PDFs.

MassSpec_PCA_DEs (R) Runs PCA on hypothalamus samples from normalized proteomics (after collapsing duplicate protein names) and saves a PC1–PC2 plot. Then uses limma to compute region- and region-specific P15 vs P14 differential expression contrasts, exports DE tables, and generates volcano plots (including WGCNA module annotations).

MassSpec_Run_WGCNA (R) Builds a signed WGCNA network on filtered proteomics data (bicor; chosen soft-threshold), writes module assignments, and correlates module eigengenes with region×age traits with BH correction. Produces a module–trait heatmap and GO barplot panels per module, exporting PDFs and summary tables.

MassSpec_Run_fGSEA (R) Runs multilevel FGSEA on a ranked hypothalamus-specific DE list using GO gene sets and exports the enrichment table. Produces a filtered, ordered NES barplot (direction-coded by NES sign) and saves it as a PDF.

Multiple_t-tests Loads a numeric CSV and performs comparisons in column pairs (1 vs 2, 3 vs 4, etc.), choosing parametric (paired/unpaired t-test) or nonparametric (Wilcoxon/Mann–Whitney) based on Shapiro–Wilk normality. Applies BH-FDR across paired tests (and uses Tukey/Dunn adjustments for unpaired cases as implemented), then exports a *_results.csv with corrected p-values and stars.

PCA_Microglia Standardizes numeric features from a single CSV, runs PCA (2D) and UMAP (2D), and clusters the scaled data with K-means (k=4). Saves an annotated CSV with embeddings and cluster labels, prints variance/loadings, computes cluster centers and distances, and plots the UMAP colored by cluster.

PCA_Microglia_Prep Scans one or more directories for specific Imaris output CSV patterns, reads each file (skipping header rows), and keeps only selected ID + measurement columns. Renames FilamentID→ID where needed, outer-merges everything into one wide table, and saves a cleaned merged CSV.

PCA_Spines Selects defined spine morphometric features, standardizes them, assigns K-means clusters (k=4), and computes PCA/UMAP embeddings with PCA loadings exported to CSV. It also classifies spines into Stubby/Filopodia/Mushroom/Thin via simple rules, then exports the annotated dataset plus per-filament and cluster×spine-type count summaries.

PCA_withPermutationTesting Concatenates all CSVs in a directory, assigns coarse clusters from the first two ID characters, and runs PCA (2D) followed by UMAP on PCA scores. Quantifies separation by computing cluster-centroid distances in PCA space and uses a feature-wise permutation test to generate null distance distributions and p-values, saving results and plotting UMAP plus observed-vs-null histograms.

Photometry Loads raw 470 nm (Ca-dependent) and 415 nm (isosbestic) traces for two fibres, subtracts 470 background, and constructs time in seconds (assumes 30 fps). Fits a bi-exponential to the isosbestic drift, linearly scales it to the calcium channel, subtracts this baseline to detrend, and exports the processed signal.

QuPath_CellCounts Cleans a QuPath/Cellfinder per-structure summary by removing excluded structures and simplifying region names by collapsing subregions into parent labels. Computes cells/mm³, optionally collapses MPOA subregions into a single MPOA row, sorts by density, and exports a compact Excel summary.