Sign in

James Lingford

@jameslingford.bsky.social
263 followers 337 following 94 posts

PhD student in structural biology with @greening.bsky.social and @knottrna.bsky.social at Monash Uni. (he/him) Interested in hydrogenases, evolution, protein design. 💻 www.jameslingford.com

PostsRepliesMedia
Reposted by James Lingford
Jess Niblo @jessniblo.bsky.social · 22/09/2026
Excited to share our manuscript “A geothermal amoeba sets a new upper temperature limit for eukaryotes” is out in Cell! This work, led by @oliverio.bsky.social & @hbrappap.bsky.social, introduces I. cascadensis, which divides at 63C, beating the prior eukaryotic limit of 60C. doi.org/10.1016/j.ce...
cell.com
A geothermal amoeba sets a new upper temperature limit for eukaryotes
The amoeba Incendiamoeba cascadensis demonstrates that eukaryotic life can withstand temperatures beyond what was thought possible and sheds light on molecular strategies for survival in extreme heat.
1123 0
Reposted by James Lingford
Martin Steinegger 🇺🇦 @martinsteinegger.bsky.social · 26/08/2026
Foldseek-Interface enables fast search/clustering of the protein interface universe! We clustered 3.1M PDB dimers into 77,167 groups and found new interfaces keep appearing even as fold discovery plateaus. 🧵 📄 www.biorxiv.org/content/10.6... 🔎 search.foldseek.com/interface 🌐 interface.foldseek.com
110941 0
James Lingford @jameslingford.bsky.social · 10/08/2026
"Enticing as it is to imagine that, by saving us time on mundane tasks, LLMs will provide us with more time to think deeply and develop projects completely, our results temper such hopes." arxiv.org/abs/2607.17397
arxiv.org
The unintended consequences of large language models as a labor-augmenting technology in science
As a labor-augmenting technology, large language models (LLMs) have the potential to accelerate scientific activity across the research pipeline. But even if LLMs perform on par with human experts at ...
020 0
Reposted by James Lingford
Yo Akiyama @yoakiyama.bsky.social · 21/07/2026
Excited to share our now published paper @cp-cell.bsky.social highlighting advances in modeling the evolution of protein-protein interactions with MSA Pairformer. Big thanks to Zhidian Zhang, Olivia Tang, @eunbelivable.bsky.social @milot.bsky.social @martinsteinegger.bsky.social and @sokrypton.org!
cell.com
Expanding the scope of protein language modeling to protein-protein interactions with MSA Pairformer
Protein language models have excelled at modeling individual proteins, but extending these capabilities to protein complexes remains a major challenge. MSA Pairformer, a parameter-efficient protein la...
26027 0
Reposted by James Lingford
Rob Edwards @linsalrob.bsky.social · 22/07/2026
Starting with the DNA sequence of a #phage genome, #PholdAPhage will create a 3D reconstruction of the complete phage, almost like you did cryoEM on it! Check out Renee's awesome software github.com/reneegreen81... to assemble unknown phage particles one protein at a time
A computer generated 3D reconstruction of a phage, based solely on its genome sequence. For this image, we started with the genome, and used PholdAPhage to create the structures. Colours are based on pLDDT scores.
39347 0
James Lingford @jameslingford.bsky.social · 17/06/2026
Something that I am very excited about: the latest version of @chimerax.ucsf.edu (v1.12) now has a cavity analysis function
0205 0
James Lingford @jameslingford.bsky.social · 04/05/2026
I care 🥲
110 0
James Lingford @jameslingford.bsky.social · 15/04/2026
Amazing paper. A megadalton enzyme complex that: - combines both electron bifurcation AND confurcation steps - has a new type of iron-sulfur cluster - a new kind of weird electron transfer path - super charges electrons to reduce aromatic rings - also powers respiration 😳
0173 0
Reposted by James Lingford
Tomas Pascoa @tomaspascoa.bsky.social · 14/04/2026
1/ Excited to share our preprint! 🥳 Degradation of aromatic compounds, including BTEX pollutants, requires highly endergonic aromatic ring reduction. Using #cryoEM and in situ #cryoET, we show how BCRII couples electron bifurcation modules in one giant redox machine www.biorxiv.org/content/10.6...
510237 0
Reposted by James Lingford
Debnath Ghosal @debnathghosal.bsky.social · 12/04/2026
One of biggest mysteries in biology: how did complex eukaryotic cells evolve from simple microbes? ~1.8 billion years ago, an archaeal cell likely merged with a bacterium to form the first eukaryotic cell, but can we ever find direct evidence of this transformative event? 🦠 🚶‍♂️
310840 0
Reposted by James Lingford
Olivia Ghosh @oliviamghosh.bsky.social · 30/03/2026
Really excited that this major work from my PhD is finally published in @plosbiology.org ! In it, we were trying to tackle a fundamental question in evolution - how do genetic mutations map onto evolutionary fitness? (1/n) journals.plos.org/plosbiology/...
journals.plos.org
Genotype-fitness mapping of adaptive mutants reveals shifting low-dimensional structure across divergent environments
Predicting the effect of a genetic mutation on fitness is a major challenge in evolutionary biology. This study uses fitness effects of a large collection of adaptive yeast mutants in multiple lab env...
28339 0
James Lingford @jameslingford.bsky.social · 28/03/2026
Found out today that @kyclark.bsky.social has made their great 'Mastering Python for Bioinformatics' book into a YouTube series. youtube.com/playlist?lis...
youtube.com
Mastering Python for Bioinformatics - YouTube
Mastering Python for Bioinformatics (O'Reilly, 2022) by Ken Youens-Clark demonstrates how to write effective Python code and how to use tests to write and re...
1269 0
Reposted by James Lingford
Martin Steinegger 🇺🇦 @martinsteinegger.bsky.social · 17/03/2026
AlphaFold database has entered the era of complexes. Together with NVIDIA, DeepMind and EBI, we use ColabFold, OpenFold and MMseqs2-GPU to predict ~31 million complexes (homo & hetro-dimers) resulting in 1.8 million high-quality predictions 📄 research.nvidia.com/labs/dbr/ass... 🌐 alphafold.ebi.ac.uk
8263110 0
Reposted by James Lingford
Kaitlin @justktln2.bsky.social · 14/03/2026
This has been up for a while but I haven’t really publicized it. Introducing ciMIST: sparse, self-consistent network models of local and global protein conformational entropy, learned from molecular dynamics. This helps with analyzing MD and connecting to experiments www.biorxiv.org/content/10.1...
23210 0
Reposted by James Lingford
Brett Baker @archaeal.bsky.social · 13/03/2026
Having a hard time keeping up with the fast-moving field of origin of eukaryotes/Asgards archaea? Our new perspective article can help, we review the status of research over the last decade and where it's going. The archaeal roots of eukaryotic life www.pnas.org/doi/10.1073/...
716976 0
Reposted by James Lingford
Mohammed AlQuraishi @moalquraishi.bsky.social · 13/03/2026
New OpenFold3 preview out! (OF3p2) It closes the gap to AlphaFold3 for most modalities. Most critically, we're releasing everything, including training sets & configs, making OF3p2 the only current AF3-based model that is functionally trainable & reproducible from scratch🧵1/9
124590 0
Reposted by James Lingford
Claus Wilke @clauswilke.com · 11/03/2026
New paper showing that much of the apparent success of protein language models in predicting mutational effects is a mirage: These models mostly memorize sites. 1/ www.biorxiv.org/content/10.6...
biorxiv.org
617972 0
Reposted by James Lingford
Rebecca Sear @rebeccasear.bsky.social · 27/02/2026
“Some have even argued that private funding is superior, maintaining that it is more flexible, less prone to groupthink, and reduces the “burden” on taxpayers. But can the private sector really replace public funding? History suggests not” www.science.org/doi/10.1126/...
science.org
Private money cannot replace public funding of science
Who should pay for American science? In the current political climate, many are looking to the private sector to compensate for cuts in public funding. At the Harvard School of Public Health—particula...
1136 0
James Lingford @jameslingford.bsky.social · 27/02/2026
A minimal PyTorch re-implementation of AlphaFold2 github.com/ChrisHayduk/... We need more stuff like this. The incentive is to always be making the new groundbreaking tool, while maintenance + improvement of "old" tools gets left behind. But there is still heaps to be gained in "old" tools
092 0
James Lingford @jameslingford.bsky.social · 26/02/2026
Aerobic syngas conversion: opportunities, challenges, and solutions www.sciencedirect.com/science/arti...
021 0
James Lingford @jameslingford.bsky.social · 26/02/2026
Any paper that quotes Italo Calvino is an instant 10/10 in my book 'An Interpretation, Survey, and Outlook of Microbial Macroecology' by @shoestrapped.bsky.social ecoevorxiv.org/repository/v...
l mondo è così complicato, aggrovigliato e sovraccarico che per vederci un po chiaro è necessario sfoltire, sfoltire.

(The world is so complicated, tangled, and overloaded that to
see into it with any clarity you must prune and prune.)

Italo Calvino Se una notte d’inverno un viaggiatore (If on a Winters Night a Traveler)
120 0
James Lingford @jameslingford.bsky.social · 24/02/2026
Insects were my first scientific love (before I knew about proteins). So I love hearing about some new weird insect "Temnothorax kinomurai is the first ant species known to lack both workers and males and to consist exclusively of queens" doi.org/10.1016/j.cu... @currentbiology.bsky.social
Temnothorax kinomurai, a workerless, parthenogenetic social parasite.
000 0
James Lingford @jameslingford.bsky.social · 23/02/2026
Their recent Science paper summarised in the review above, here: www.science.org/doi/full/10....
science.org
Conformational ensembles reveal the origins of serine protease catalysis
Enzymes exist in ensembles of states that encode the energetics underlying their catalysis. Conformational ensembles built from 1231 structures of 17 serine proteases revealed atomic-level changes acr...
010 0
James Lingford @jameslingford.bsky.social · 23/02/2026
Understanding how enzymes work: the journey to ensemble–function studies febs.onlinelibrary.wiley.com/doi/10.1111/... This is an exceptionally well written review on how enzymes work. Rather than the structure-function relationship, Herschlag & Du analyse the ensemble-function relationship
193 0
James Lingford @jameslingford.bsky.social · 22/02/2026
Structural ontogeny of protein-protein interactions: www.science.org/doi/10.1126/... Fundamental work on how PPIs (co)evolve, combining directed evolution and machine learning to reveal the role of chance and contingency.
0136 0
Reposted by James Lingford
Yun S. Song @yun-s-song.bsky.social · 21/02/2026
Can we simulate realistic evolutionary trajectories and “replay the tape of life”? In this work, we propose a flexible, generalizable deep learning framework for modeling how the entire protein sequence evolves over time while capturing complex interactions across sites. 1/n doi.org/10.64898/202...
doi.org
38735 0
Reposted by James Lingford
Pedro Leão @pedroleao.bsky.social · 20/02/2026
You go, Katy! 🚀 I’m so proud of you, and so happy to finally see this in its final form! Count me in on your corner to keep on cheering for you! Looking forward to doing more amazing science together 💪 Check @katyappler.bsky.social thread below for highlights and the full paper here: rdcu.be/e4A70
nature.com
Oxygen metabolism in descendants of the archaeal-eukaryotic ancestor - Nature
Sequencing of marine sediments finds 136 newly identified Heimdallarchaeia and several novel lineages, and indicates that Heimdallarchaeia evolved distinct metabolic capabilities from other Asgar...
162 0
James Lingford @jameslingford.bsky.social · 20/02/2026
Asgard archaeal origin of microtubules: www.biorxiv.org/content/10.6...
073 0
James Lingford @jameslingford.bsky.social · 20/02/2026
I present this extremely niche meme for the hydrogenase community
"The SuiHyd Squad"
141 0
James Lingford @jameslingford.bsky.social · 20/02/2026
Totally agree. Lots of assumptions baked in there
000 0
James Lingford @jameslingford.bsky.social · 19/02/2026
"GenAI increases output and helps programmers expand into new domains—but only for senior- level developers. Early-career developers, despite being the most enthusiastic adopters, see no measurable gains" www.science.org/doi/10.1126/...
science.org
Who is using AI to code? Global diffusion and impact of generative AI
Generative coding tools promise big productivity gains, but uneven uptake could widen skill and income gaps. We train a neural classifier to spot artificial intelligence (AI)–generated Python function...
122 0
James Lingford @jameslingford.bsky.social · 19/02/2026
Cross-post on LinkedIn, since everyone is on LinkedIn these days: www.linkedin.com/posts/jamesl...
linkedin.com
Very excited to share our work published today in Nature: ‘Oxygen metabolism in descendants of the archaeal-eukaryotic ancestor’. This was a huge effort led by the amazing Kathryn Appler from Brett…...
Very excited to share our work published today in Nature: ‘Oxygen metabolism in descendants of the archaeal-eukaryotic ancestor’. This was a huge effort led by the amazing Kathryn Appler from Brett ...
131 0
James Lingford @jameslingford.bsky.social · 19/02/2026
Thanks Kate 😊
010 0
James Lingford @jameslingford.bsky.social · 19/02/2026
We found [NiFe]-hydrogenase complexes with structural architectures very similar to that of respiratory complex I. It's been known for a long time that [NiFe]-hydrogenase and complex I are evolutionarily related, but prior work showed that the configuration of their subunits differed (e.g., MBH)
Overview of the modular architecture of complex I–like respiratory complexes, from the fantastic review by Yu et al. (2021) JBC: https://www.jbc.org/article/S0021-9258(21)00529-9/fulltext
161 0
James Lingford @jameslingford.bsky.social · 19/02/2026
When we looked into the what sort of multimeric enzyme complexes these hydrogenases were predicted to make, we were amazed to see a menagerie of many new forms that were unlike anything we'd seen before (to our knowledge)
Supplementary Fig. 46 | Overall structural architecture and shared subunit homology in Complex I-like  and MBH-like complexes. AF2 multimer models are divided into structures that share a Complex I-like or  an  MBH-like  structural  architecture.  For  MBH-like  complexes,  the  P.  furiosus membrane-bound  hydrogenase, MBH  (PDB  ID:  6CFW)26,  is  shown  for  comparison.  The  order  and  orientation  of  the  membrane-arm subunits relative to the peripheral-arm subunits is characteristic of MBH and distinguishes  those structures from Complex I-like complexes26,88. For Complex I-like complexes, the E. coli Complex I  (PDB ID: 7P62)94 is shown for comparison. The order and orientation of membrane-arm subunits relative  to the peripheral-arm subunits in these AF2 models is characteristic of Complex I. In other words, when all  homologous antiporter subunits are viewed in the same register, the peripheral-arm subunits in MBH-like  complexes  will  appear  rotated  at  ~180°  relative  to  Complex  I-like  peripheral-arm  subunits.  We  have  highlighted areas on our AF2 models where expected subunits are absent. Missing subunits may be a result  of  our  annotation  step  failing  to  capture  all  relevant  sequences  for  AF2  input.  A  table  showing  subunit  names and their corresponding coloring in protein models is shown as a color key. Homologous subunits  share the same coloring. The colors of different subunits are consistent across this manuscript unless stated  otherwise.  Acronyms  for  the  complexes,  removing  the  taxonomic  suffix  and  adding  numbering  Kari-1  (GCA_015523565), Kari-2 (GCA_024280435), Njord-1 (M288), Njord-2 (D4998_C1112_H3_Bin_254), Hod-1  (M3_38_Bin_455),  Hod-2  (GCA_024276395),  Hod-3  (M3_30_Bin_130), Gerd  (GCA_016839405),  Asgard-1  (AB3033_2), and Asgard-2 (GCA_021162905).Supplementary Fig. 29 | CombFold models of putative electron bifurcating complexes. a. The putative  Nuo(EFG)
2-Mvh(ADG)
2-Etf(AB)
2-Hdr
2 electron  bifurcating  complex  found  in  Hodarchaeales MAG  GCA_016840025. Top: individual AF2 models of the NuoEFG, MvhADG, and EtfAB-Hdr modules forming  part  of  the  total  complex.  Bottom: the  CombFold  assembled  model  shown  from  three  different  views.  Subunits  are  colored  according  to  the  labelling  shown  in  the  top  panel. b. The  putative  Nuo(EFG)4- Frh(AG)4  electron  bifurcating  complex  found  in Hodarchaeales MAG  GCA_020353515.  Top:  individual  AF2 models of the NuoEFG and FrhAG modules forming part of the total complex. Bottom: the CombFold  assembled model shown from three different views. Subunits are colored according to the labelling shown  in the top panel. Frh: F
420-reducing hydrogenase (a [NiFe]-hydrogenase). Mvh: Methyl viologen-reducing  hydrogenase  (a  [NiFe]-hydrogenase).  Etf:  electron  transferring  flavoprotein.  Hdr:  heterodisulfide  reductase.
140 0
James Lingford @jameslingford.bsky.social · 19/02/2026
One particular protein family we focussed on were the [NiFe]-hydrogenases, which are ancient metalloenzymes that let microbes metabolise hydrogen gas (H2). @katyappler.bsky.social noticed that the Asgard have a lot of diverse hydrogenase sequences that we hadn't seen before.
Supplementary Fig. 44 | AlphaFold2 monomer models of [NiFe]-hydrogenase large subunits Group 4.  Maximum likelihood phylogeny was inferred from a supermatrix of 3007 sequences and 440 amino acid  positions (from 2216 initial positions), displaying the diversity of Asgard archaea hydrogenases collapsed  to highlight [NiFe]-Groups 4 (blue) identified by the external ring. The tree was generated using IQ-TREE  v2.0.7 (1000 ultrafast bootstrap replicates and calculation of 1000 SH-alRT) with Q.pfam+C50+R8 best-fit  model  chosen  according  to  Bayesian  Information  Criterion  (BIC).  Previously  identified  subgroups  are  labeled in black. The novel hydrogenase groups identified and named in this study are distinguished by  branch and label color-coded according to the group. The black check marks highlight the 14 large catalytic  subunits chosen for monomer modeling for subgroups 4j, 4l, 4p, and 4q. Support values are shown when  ultrafast (UF) bootstrap support is greater than or equal to 95 and SH-aLRT support is greater than or equal  to 80. Individual monomers colored by the predicted local distance difference test (pLDDT) scores. Only  top ranked AF2 models are shown. The phylogeny was visualized with iTOL and midpoint rooted between  the known hydrogenase clades and pruned to show each group, separately.Supplementary Fig. 43 | AlphaFold2 monomer models of [NiFe]-hydrogenase large subunits Group 3.  Maximum likelihood phylogeny was inferred from a supermatrix of 3007 sequences and 440 amino acid  positions (from 2216 initial positions), displaying the diversity of Asgard archaea hydrogenases collapsed  to  highlight  [NiFe]-Groups  3  (orange)  identified  by  the  external  ring. The  tree  was  generated  using  IQ- TREE  v2.0.7  (1000  ultrafast  bootstrap  replicates  and  calculation  of  1000  SH-alRT)  with  Q.pfam+C50+R8  best-fit model chosen according to Bayesian Information Criterion (BIC). Previously identified subgroups  are labeled in black. The novel hydrogenase groups identified and named in this study are distinguished  by  branch  and  label  color-coded  according  to  the  group.  The  black  check  marks  highlight  the  12  large  catalytic subunits chosen for monomer modeling for subgroups 3a, 3b, 3e, 3f, and 3k. Support values are  shown when ultrafast (UF) bootstrap support is greater than or equal to 95 and SH-aLRT support is greater  than or equal to 80. Individual monomers colored by the predicted local distance difference test (pLDDT) scores. Only top ranked AF2 models are shown. The phylogeny was visualized with iTOL and midpoint  rooted between the known hydrogenase clades and pruned to show each group, separately.
130 0
James Lingford @jameslingford.bsky.social · 19/02/2026
Since my background is in protein structure, one facet of this project that I find extremely interesting is that Asgard archaea contain weird and wonderful proteins that have never been seen before. We used AlphaFold to predict and explore the structures of these Asgard proteins
Fig. 3 | Asgard archaea produce structurally diverse respiratory complexes. a, Phylogeny of hydrogen-evolving Group 4 [NiFe]-hydrogenases, displaying the novel subclades 4j–r with blue branches and the Asgard lineage in the second ring. AlphaFold2 Multimer structures outline the phylogeny, indicating a transition from an MBH-like to a Complex I-like structure. Asgardarchaeia structures from 4k and 4l form an MBH-like complex represented by cryogenic electron microscopy(cryo-EM)-validated Pyrococcus furiosus MBH (Protein Data Bank (PDB): 6CFW) and membrane-bound sulfane sulfur reductase (MBS) (PDB: 6U8Y) structures. The other complexes, 4m, 4p and 4q, have structures that are more similar to the E. coli Complex I (PDB: 7P62).
192 0
James Lingford @jameslingford.bsky.social · 19/02/2026
This has very exciting implications for eukaryogenesis. The first eukaryotes arose during the 'great oxygenation event' (a time of great upheaval in the Earth's history). Our findings suggest that the Asgard/eukaryote ancestor may have benefited from this rise in oxygen through aerobic respiration.
120 0
James Lingford @jameslingford.bsky.social · 19/02/2026
@katyappler.bsky.social and @archaeal.bsky.social went and hunted down new Asgard archaea across the globe. They ~doubled the previously known genomic catalogue of Asgard archaea. From this treasure trove of new genomic information, we infer they have oxygen metabolism.
120 0
James Lingford @jameslingford.bsky.social · 19/02/2026
The Asgard archaea are the closest living relatives to eukaryotes, and can therefore tell us a lot about one of the most pivotal events in the history of evolution: the origin of eukaryotes.
120 0
James Lingford @jameslingford.bsky.social · 19/02/2026
Also, Carl Zimmer wrote a short story about this work! Great to see basic exploratory biological research like this get highlighted by such a huge name. Inspiring stuff. bsky.app/profile/carl...
120 0
James Lingford @jameslingford.bsky.social · 19/02/2026
Our work is published today: ‘Oxygen metabolism in descendants of the archaeal-eukaryotic ancestor’. This was a huge effort lead by @katyappler.bsky.social. Extremely grateful to have been a part of this amazing project! 😊🦠🧬 Links: www.nature.com/articles/s41... www.nature.com/articles/s41...
Eukaryogenesis in light of an expanded catalogue of Asgard genomes. a, Simplified, scaled timeline spanning from before the Last Asgard archaea Common Ancestor (LAsCA) to today. Thin bands mark predicted time ranges  of relevant events (for example, GOE), thicker bands represent processes  (for example, eukaryogenesis), and brackets indicate the period shown in b. The timeline further highlights milestones, including potential early eukaryotic fossils60 and the modern-day co-occurrence of Heimdallarchaeia and Alphaproteobacteria observed in this study (interaction likely originated earlier). Fig. 1 | Expanded genomic diversity of Asgard archaea. a, Maximum-likelihood phylogeny based on 47 non-ribosomal markers (NM47)using the WAG + C10 + R4 model with 100 nonparametric bootstrap pseudoreplicates, including 869 Asgardarchaeota MAGs and 309 outgroup genomes. The blue branches (lower right) indicate the new Asgardarchaeota classes, Ranarchaeia, and the recently proposed Asgardarchaeia4. The concentric rings denote (in to out): the predicted genome size, metabolic guilds based on Pfam clustering, sampling locations, and black stars on the outside mark MAGs added by this study. Asgard, Asgardarchaeia; Atabey, Atabeyarchaeia; Baldr, Baldrarchaeia; Frey/Jord,  Frey/Jordarchaeia; Gerd, Gerdarchaeales; Heimdall, Heimdallarchaeaceae;  Hel, Helarchaeales; Hermod, Hermodarchaeia; Hod, Hodarchaeales;  Kari, Kariarchaeaceae; Loki, Lokiarchaeales; Njord, Njordarchaeales;  Odin, Odinarchaeia; Ran, Ranarchaeia; Sif, Sifarchaeia; Thor, Thorarchaeia;  Wukong, Wukongarchaeia. b, SR4-recoded phylogeny of the same genome  set inferred with the model GTR + C60 + G and 100 nonparametric bootstrap pseudoreplicates (Methods). This updated catalogue constitutes a large increase in the medium- to high-quality publicly available genomes (completeness >50% and contamination and redundancy <10%) with 65.3% from the Guaymas Basin and 34.7% from the Bohai Sea. The encircled numbers represent MAGS added by this study. The scale bars in bothsubpanels represent the average number of substitutions per site.Map created in BioRender; Appler, K. https://biorender.com/147ieoc(2025).
34818 0
Reposted by James Lingford
Thijs J. G. Ettema 🦠🔬🇳🇱🇸🇪🇪🇺 @ettema.bsky.social · 18/02/2026
Finally out in @nature.com: a new piece of the puzzle of how complex life evolved. Lead by @archaeal.bsky.social & @katyappler.bsky.social. Great collab with @greening.bsky.social and @kassipan.bsky.social. More pieces to follow soon! www.nature.com/articles/s41...
48530 0
Reposted by James Lingford
Carl Zimmer @carlzimmer.com · 18/02/2026
One of the biggest questions in biology is how complex cells evolved about 2 billion years ago. Here's my new story on how scientists are solving the mystery of eukaryotes like us. Gift link: nyti.ms/4qMbo22
A pilot in a submersible vehicle collecting sediments samples in 30 meters of water looking for Asgards (microbial relatives of eukaryotes)
9342105 0
Reposted by James Lingford
Kathryn (Katy) Appler @katyappler.bsky.social · 18/02/2026
At long last! Check out the link to our publication in @nature.com to learn more. doi.org/10.1038/s415... 12/12
doi.org
Oxygen metabolism in descendants of the archaeal-eukaryotic ancestor - Nature
Sequencing of marine sediments finds 136 newly identified Heimdallarchaeia and several novel lineages, and indicates that Heimdallarchaeia evolved distinct metabolic capabilities from other&nbsp;Asgar...
1229 0
Reposted by James Lingford
Brett Baker @archaeal.bsky.social · 18/02/2026
New paper from my team detailing a greatly expanded genomic database of Asgard archaea revealing of high energy metabolism those related to eukaryotes! Led by @katyappler.bsky.social lots of help from @jameslingford.bsky.social @valdeanda.bsky.social @kassipan.bsky.social doi.org/10.1038/s415...
1015161 0
James Lingford @jameslingford.bsky.social · 18/02/2026
AI can of course help with coding problems as a sprt of better Google search (it's helped me), but outsourcing all work to an AI sounds like a Faustian bargain. A AI companies are making huge promises about how all jobs can be outsourced to an AI, but of course they say that, they want money.
010 0
James Lingford @jameslingford.bsky.social · 18/02/2026
And on top of this, these AI agents can cost hundreds of dollars per month in subscriptions. At this point, wouldn't it be both easier, more productive, and intellectually fulfilling to just do "non-vibe" coding? 2/3
100 0
James Lingford @jameslingford.bsky.social · 18/02/2026
Keep hearing hype around vibe coding and how its going to replace all coding very soon. So I look into pro-vibe coding communities on reddit. They all say you have to hold the hand of the AI at every step of the way and have very long but specific prompts, otherwise the AI makes a mess... 1/2
120 0
Reposted by James Lingford
Charles Bayly-Jones @charlesbj.bsky.social · 18/02/2026
I’ve officially started my lab at Monash University! 🎉 We’re diving into #cellgrowth and #lysosome biology, through the lens of molecular #structure and mechanism. I’m thrilled to say that PhD student scholarships are available—so please share widely and get in touch if you’re interested. 🤩
43616 0