Background Reproducibility and cross-species translation using the domestic pig (Sus scrofa) are limited by the lack of a standardised molecular framework for biological maturation: the pig’s developmental tempo differs substantially from the human’s, yet no tissue-resolved transcriptomic staging system exists. Synchronising porcine and human maturation is essential to move preclinical research from descriptive to predictive. Results We built a transcriptomic atlas of porcine development across five tissues (muscle, brain, liver, blood, lung) from 1,924 PigGTEx RNA-seq profiles. A single partial-least-squares (PLS) regressor staged each tissue at its native ordinal resolution and also drove cross-species transfer and biomarker extraction. Our central result: transfer of a pig-trained developmental score to other species is dominated by tissue identity, with phylogenetic distance a weaker secondary effect. Projected onto the seven-species Cardoso-Moreira atlas, transfer was strongest for brain and heart (<InlineEquation ID="IEq1"><EquationSource Format="TEX">\(\rho = 0.89\)</EquationSource></InlineEquation>–0.92) and weakest for the labile liver and ovary (<InlineEquation ID="IEq2"><EquationSource Format="TEX">\(\rho = 0.49\)</EquationSource></InlineEquation> and 0.42). Brain and heart stayed high even from pig to chicken (<InlineEquation ID="IEq3"><EquationSource Format="TEX">\(\rho \ge 0.95\)</EquationSource></InlineEquation> across <InlineEquation ID="IEq4"><EquationSource Format="TEX">\(\sim\)</EquationSource></InlineEquation>320 Myr), whereas liver and ovary collapsed. A variance partition confirmed the ranking (organ Type-II <InlineEquation ID="IEq5"><EquationSource Format="TEX">\(F = 9.0\)</EquationSource></InlineEquation>, <InlineEquation ID="IEq6"><EquationSource Format="TEX">\(p = 2\times 10^{-5}\)</EquationSource></InlineEquation>; phylogeny significant only at the species level, Spearman <InlineEquation ID="IEq7"><EquationSource Format="TEX">\(-0.76\)</EquationSource></InlineEquation>, <InlineEquation ID="IEq8"><EquationSource Format="TEX">\(p = 0.049\)</EquationSource></InlineEquation>, <InlineEquation ID="IEq9"><EquationSource Format="TEX">\(n = 7\)</EquationSource></InlineEquation>). The score independently confirmed lung, muscle, and adipose on the human dGTEx resource (<InlineEquation ID="IEq10"><EquationSource Format="TEX">\(\rho = 0.59\)</EquationSource></InlineEquation>–0.67) against an ortholog-scramble null and three controls. Because it is fit on pig and projected onto foreign-batch atlases, the transfer cannot arise from a pig-side artefact. A web application is available at <a href="https://pigdevstage.streamlit.app">https://pigdevstage.streamlit.app</a>. Conclusions The porcine developmental programme transfers to humans and more distant amniotes in a tissue-dependent manner, so the pig’s value as a developmental model is set by how conserved each organ’s developmental logic is, not by phylogenetic proximity. Validated on foreign-species atlases, the transfer sidesteps the stage–study confound that bounds within-pig staging to muscle and liver. The atlas is therefore best used organ by organ: strong for conserved organs (brain, heart) and weak for labile ones (liver).

Tissue identity is the dominant determinant of cross-species transferability of a porcine developmental programme

期刊论文
出版模式：

开放获取
发表日期：

2026年7月4日

查看全文

Tianyuan Liu,
Rujing Lei,
Ilyas M. Khan,
Peter Theobald

摘要

Background

Reproducibility and cross-species translation using the domestic pig (Sus scrofa) are limited by the lack of a standardised molecular framework for biological maturation: the pig’s developmental tempo differs substantially from the human’s, yet no tissue-resolved transcriptomic staging system exists. Synchronising porcine and human maturation is essential to move preclinical research from descriptive to predictive.

Results

We built a transcriptomic atlas of porcine development across five tissues (muscle, brain, liver, blood, lung) from 1,924 PigGTEx RNA-seq profiles. A single partial-least-squares (PLS) regressor staged each tissue at its native ordinal resolution and also drove cross-species transfer and biomarker extraction. Our central result: transfer of a pig-trained developmental score to other species is dominated by tissue identity, with phylogenetic distance a weaker secondary effect. Projected onto the seven-species Cardoso-Moreira atlas, transfer was strongest for brain and heart (\(\rho = 0.89\)–0.92) and weakest for the labile liver and ovary (\(\rho = 0.49\) and 0.42). Brain and heart stayed high even from pig to chicken (\(\rho \ge 0.95\) across \(\sim\)320 Myr), whereas liver and ovary collapsed. A variance partition confirmed the ranking (organ Type-II \(F = 9.0\), \(p = 2\times 10^{-5}\); phylogeny significant only at the species level, Spearman \(-0.76\), \(p = 0.049\), \(n = 7\)). The score independently confirmed lung, muscle, and adipose on the human dGTEx resource (\(\rho = 0.59\)–0.67) against an ortholog-scramble null and three controls. Because it is fit on pig and projected onto foreign-batch atlases, the transfer cannot arise from a pig-side artefact. A web application is available at https://pigdevstage.streamlit.app.

Conclusions

The porcine developmental programme transfers to humans and more distant amniotes in a tissue-dependent manner, so the pig’s value as a developmental model is set by how conserved each organ’s developmental logic is, not by phylogenetic proximity. Validated on foreign-species atlases, the transfer sidesteps the stage–study confound that bounds within-pig staging to muscle and liver. The atlas is therefore best used organ by organ: strong for conserved organs (brain, heart) and weak for labile ones (liver).