Pan-Homo

Paleontological and Molecular Divergence of the last common ancestor between hominins and panins

This repository provides a reproducible pipeline for the meta-analysis of molecular divergence estimates between humans and chimpanzees, evaluated against the late Miocene and early Pliocene fossil record.

---

📂 Project Structure

• data/ Contains HomoPanDivergences.csv, the primary dataset of sampled molecular estimates.

• R/ The core logic. Includes data preparation (prepare_data.R), utility functions (functions_utility.R), and the Bayesian Meta-Analysis engine (analysis_meta.R).

• models/ Stores the .rds files for the 5 Bayesian models (Full, A, B, C, and Genomic).

• figures/ Scripts and outputs for the Main Manuscript figures. Output images are located in: figures/output/ (Figure_1.png to Figure_5.png)

• supplementary/ Scripts and outputs for the Supporting Information (SI) figures and tables.

---

🚀 How to Reproduce the Analysis

Run the scripts in the following order to ensure all models are calculated before plotting:

1. Initialize and Run Models

source("R/prepare_data.R")
source("R/analysis_meta.R")

2. Generate Main Figures

source("figures/make_main_figures.R")

3. Generate SI Figures and Tables

source("supplementary/make_si_figures.R")
source("supplementary/make_si_tables.R")

---

📊 Manuscript Figures & Legends

Main Figures

Figure 1 – Molecular estimates histogram

Dashed vertical lines represent the fossil thresholds. Note that instead of following a normal distribution, studies seem to cluster in excess around important fossil discoveries.

---

Figure 2 – Interquartile range boxplots

Divergence estimates filtered by different fossil thresholds. All boxplots fit within the late Miocene (11.6–5.3 Ma).

---

Figure 3 – Full-dataset model trend

Fitting the sample of Panini/Hominini split estimates by date of publication.

---

Figure 4 – Linear regressions by source

Linear regressions for different types of source datasets. A) Includes early phenetic studies (immunological, DNA-DNA hybridization, etc.), but also later studies with virogenes and RNA; B) Mitochondrial datasets from early studies with selected sequences to later complete mitochondrial genome studies; C) Split estimates based on nuclear DNA sequences; and D) Large phylogenomic studies (also includes ‘total evidence dating’, i.e., integrative models).

---

Figure 5 – Meta-analysis of phylogenomic estimates

Forest plot depicting the specific effect size and sampled posterior distribution of genome-based studies.

---

📎 Supplementary Figures

• Fig. S1 – Density curves Min–estimate–max triplets sampled from the literature. Dashed vertical bars represent median values.

• Figs. S2–S5 – Sensitivity BMAs Full dataset and subsets filtered by 4.4 Ma, 6.2 Ma, and 7.2 Ma thresholds.

---

🛠 Requirements

R environment with the following packages:

• brms
• Stan
• tidybayes
• ggplot2
• ggdist
• ggridges
• cowplot
• sjPlot
• psych
• dplyr
• glue
• forcats
• stringr
• zoo
• scales

---

📬 Contact

João d'Oliveira Coelho joaocoelho@gorongosa.net

---