The larger catalogue for the whole series is on the parent page, the origin of life: the literature. Working notes are in
abiogenesis/research/minimal-cell-and-luca.md.
Sources cited
| Source | What it is | Access | Direction |
|---|---|---|---|
| Giovannoni et al. 2005, Science 309:1242 | Pelagibacter ubique: the smallest genome known for a free-living organism, with complete pathways for all twenty amino acids. | Abstract from PubMed | complicates |
| Nakabachi et al. 2006 | Carsonella ruddii: 159,662 bp, 182 open reading frames, and a gene inventory the authors call insufficient for essential bacterial processes. | Opened | complicates |
| Bennett & Moran 2013 | Nasuia: 112 kb, no oxidative phosphorylation. | Opened | complicates |
| Hutchison et al. 2016, Science 351:aad6253 | JCVI-syn3.0: 531,560 bp, 473 genes, 149 of unknown function, with the environment caveat in its own discussion. | Opened in full | complicates |
| Breuer et al. 2019, eLife 8:e36842 | syn3A: 452 coding genes, 91 of unclear function, of which 30 are essential. | Opened in full (CC BY) | complicates |
| Pelletier et al. 2020 | Seven genes required together to restore normal division, five of unknown function. | Opened (CC0 preprint) | complicates |
| Moger-Reischer et al. 2023, Nature 620:122 | Minimisation costs >50% fitness, regained in 2,000 generations; the minimal cell then evolves 39% faster. | Opened in full (CC BY) | undercuts |
| Kilinc, Jia & Jernigan 2025 | Protein language models assign annotated homologs to 93% of the minimal genome. | Opened | undercuts |
| Michalik et al. 2026, Nat Commun 17:2473 | Vidania at 50-52 kb, "further blurring the bacteria-organelle boundary." Post-cutoff. | Opened | complicates |
| Gianni et al. 2026, Science 391:1022 | The other end of the gap: a 45-nucleotide ribozyme at 0.2% yield over 72 days. Post-cutoff. | Opened in full (CC BY) | complicates |
Considered and not used
- Pelletier et al.'s published Cell version (2021). Only the CC0 preprint was opened; the quotations come from it and the footnote says so.
- Karr, Sanghvi & Covert 2012, Cell 150:389, the whole-cell computational model of Mycoplasma genitalium. Not on disk and not used. It measures the same organism from the opposite direction and belongs in a future revision.
- Hutchison et al. 1999, the transposon study giving "265 to 350 of the 480 protein-coding genes" as essential. Handled in the parent article, where it matters because a widely repeated 256 figure is a transposition of 265.
- The gene-count drift between papers. Hutchison gives syn3.0 as 473 (438 protein plus 35 RNA); Breuer gives syn3A as 493 with 452 coding; a 2025 methods chapter says "473, with 434 of these coding." The article attaches each number to its paper rather than reconciling them.
Open questions
- No experimental characterisation of the syn3.0 unknown genes was found. Kilinc and colleagues supply homology, which is not function.
- Whether anyone has measured how far syn3A can be reduced further, given Breuer's own suggestions for additional removals.
- The lower bound for a first cell is unknown and probably unknowable by this method, since every organism available to measure is a modern one.