iiinova — where your digital self keeps evolving

Reposted by Jordan Dorrell

CCDC Cambridge @ccdc.cam.ac.uk · 01/10/2026

part 2: You can search with a control file that is, for example "asymmetric unit components: 1.99 3.01" which would find structures with 2 or 3 components in the asymmetric unit.

012

Reposted by Jordan Dorrell

CCDC Cambridge @ccdc.cam.ac.uk · 01/10/2026

A new update from the team: We have a CSD Python API script that allows you to search using Z'', the number of chemical components in the asymmetric unit. This is contained within the Refcodes with properties script and is available in our script repository.

111

Reposted by Jordan Dorrell

Chemistry World @chemistryworld.com · 14/04/2026

Growing crystals has always been half science & half art, and the battle between polymorphs involves the competing drivers of thermodynamic stability & kinetic control. @rachelbrazil.bsky.social www.chemistryworld.com/features/the...

chemistryworld.com

The curious case of disappearing polymorphs

From crystallographers' beards to billion-dollar lawsuits, the phenomenon of disappearing polymorphs has puzzled chemists for decades. Now researchers are finally unravelling the science behind why st...

061

Reposted by Jordan Dorrell

Ganna (Anya) Gryn’ova 🇺🇦 @grynova.bsky.social · 12/03/2026

We have a fully funded PhD position (UK students only) at @uobchemistry.bsky.social to work on #multiscale #modelling and #ML of organic #materials! Details www.grynova-ccc.org/jobs.html #chemsky #PhDjobs #chemPhD Please spread the word!

0149

Reposted by Jordan Dorrell

Alchemy @aichemyhub.bsky.social · 06/02/2026

📢 We’re hiring! Join us as a Postdoctoral Research Associate in AI/ML-Driven Chemical Discovery Work at the intersection of AI, machine learning & chemistry on cutting-edge research. 📍 Liverpool | ⏰ Apply by 20 Feb 2026 👉

buff.ly

Postdoctoral Research Associate in AI/Ml-Driven Chemical Discovery - Grade 7 at University of Liverpool

Apply now for the Postdoctoral Research Associate in AI/Ml-Driven Chemical Discovery - Grade 7 role on jobs.ac.uk - the leading job board for higher education jobs. View details.

043

Reposted by Jordan Dorrell

Andy Cooper @aicooper.bsky.social · 05/02/2026

Please repost if you’re in the area of #AI or #chemistry. Funded by EPSRC. #chemsky

11210

Jordan Dorrell @stochasticchemist.bsky.social · 27/01/2026

"a "Hive Mind" method that connects autonomous laboratory robots with human scientists and AI agents". Sounds exciting! Will you have multiple people per hive mind or just one human with a flock (what's the collective noun?) of robots/AIs?

100

Reposted by Jordan Dorrell

Rachel Crespo Otero @rcrespootero.bsky.social · 19/01/2026

I have a three-year PDRA position to work on the development of adaptive embedding methods. Please share this with anyone who may be interested. #CompChem www.ucl.ac.uk/work-at-ucl/...

ucl.ac.uk

UCL – University College London

UCL is consistently ranked as one of the top ten universities in the world (QS World University Rankings 2010-2022) and is No.2 in the UK for research power (Research Excellence Framework 2021).

046

Jordan Dorrell @stochasticchemist.bsky.social · 08/01/2026

If we switch to GUIs, the experimentalists might realise how easy we have it!

220

Jordan Dorrell @stochasticchemist.bsky.social · 16/12/2025

Super excited to get this preprint online with @graemeday.bsky.social ! With SAUCE, we significantly improve on the efficiency of random structure searching for crystals containing multiple symmetry-independent molecules! chemrxiv.org/engage/chemr... #compchem #csp

142

Jordan Dorrell @stochasticchemist.bsky.social · 10/12/2025

If I've misunderstood anything, I do apologize. Please correct me in the thread. Overall I think the paper's going to do very well and I look forward to reading the final print! 12/12

000

Jordan Dorrell @stochasticchemist.bsky.social · 10/12/2025

reduce the number of force evaluations required per optimisation, then their method is likely to be overall cheaper than contemporary CSP methods and I would expect to see the model as a strong contender in the next CSP blind test. 11/12

100

Jordan Dorrell @stochasticchemist.bsky.social · 10/12/2025

There is one final point in which I think that they're probably *under*selling their work. The accuracy of their generated structures is likely to improve if optimised with DFT. If their model can reduce the number of structures that need to be optimised, and 10/12

100

Jordan Dorrell @stochasticchemist.bsky.social · 10/12/2025

I'd really like to see how the model compares to the blind test participants using the 7th blind test's criteria for a match. I would also be really interested to see how Sol_c changes with the number of crystals generated and also the number of symmetry-independent molecules. 9/12

100

Jordan Dorrell @stochasticchemist.bsky.social · 10/12/2025

They do summarise the performance of each group for every target in their SI but they haven't reported the performance of OXtal on each target so we can't compare. 8/12

100

Jordan Dorrell @stochasticchemist.bsky.social · 10/12/2025

If they're averaging over each group's performance, that's flawed in the other way because the best performing groups/methods are impeded my the lower performing groups/methods. 7/12

100

Jordan Dorrell @stochasticchemist.bsky.social · 10/12/2025

If they're counting a successful prediction by any group as a success for "DFT-based submissions" then they're probably being overly generous to DFT because they're sampling across an ensemble of different DFT approaches. 6/12

110

Jordan Dorrell @stochasticchemist.bsky.social · 10/12/2025

They compare their model initially to other generative models on CSD structures but later compare their performance against a "aggregate of reported expensive DFT-based submissions" for previous CSP blind tests. I'm not entirely sure how they're aggregated though. 5/12

100

Jordan Dorrell @stochasticchemist.bsky.social · 10/12/2025

In the preprint, the criteria for "approximately solved" (Sol_c) was an RMSD of less than 2 Å for a cluster of 15 molecules (a significantly looser tolerance). They're not using the word "match" here, but it really feels implied that that's what they have. 4/12

100

Jordan Dorrell @stochasticchemist.bsky.social · 10/12/2025

The authors use the @ccdc.cam.ac.uk's COMPACK metric to overlay crystals and minimise the RMSD between clusters of N molecules. COMPACK is also used in the CSP blind tests but in the latest blind test, the criteria for a match was an RMSD of less than 1 Å for a cluster of 30 molecules. 3/12

120

Jordan Dorrell @stochasticchemist.bsky.social · 10/12/2025

Unfortunately, the data as presented makes me feel that the claims made by the authors are somewhat exaggerated. I'm particularly concerned with the metrics which are described as "standard CSP metrics". 2/12

110

Jordan Dorrell @stochasticchemist.bsky.social · 10/12/2025

"OXTAL: AN ALL-ATOM DIFFUSION MODEL FOR ORGANIC CRYSTAL STRUCTURE PREDICTION" was put into preprint (arXiv:2512.06987) this weekend! We've seen some decent generative models for inorganic CSP but this model marks a big step forward for generative models for organic crystals! 1/12 #compchem

120

Reposted by Jordan Dorrell

Graeme Day @graemeday.bsky.social · 27/11/2025

I'm excited to share the latest paper from our team, just published in Nature Communications: rdcu.be/eRTSs "Exploring organic chemical space for materials discovery using crystal structure prediction-informed evolutionary optimisation" #compchemsky #chemsky 1/n

Schematic of an evolutionary algorithm for generating new organic molecules, with crystal structure prediction integrated into the fitness function calculation.

1135

Jordan Dorrell @stochasticchemist.bsky.social · 05/10/2025

New addition for the Day group office?

010

Jordan Dorrell @stochasticchemist.bsky.social · 29/09/2025

Choose comp-chem and you won't even have to go into the office!

220

Reposted by Jordan Dorrell

Graeme Day @graemeday.bsky.social · 24/09/2025

Just about ready for our first workshop on mol-cspy: our source software for crystal structure prediction gitlab.com/mol-cspy/mol... A massive thank you to the research team in getting material together for this. #compchemsky #chemsky

1125

Reposted by Jordan Dorrell

Mayor of London, Sadiq Khan @london.gov.uk · 22/09/2025

Thousands of Londoners have indefinite leave to remain. They have legal rights and are our friends, neighbours and colleagues, contributing hugely to our city. Threatening to deport people living and working here legally is unacceptable.

theguardian.com

Farage vows to scrap indefinite leave to remain, placing thousands at risk of deportation

Reform UK plans to force non-citizens to apply for visas with high salary thresholds and no access to NHS services

1413012896

Jordan Dorrell @stochasticchemist.bsky.social · 19/09/2025

Strange way to ask what sector of science people work in.

000

Jordan Dorrell @stochasticchemist.bsky.social · 19/09/2025

I had a fantastic time at the #MMMHub conference this week. I had some really great conversations! Thank you to all the attendees who voted me runner up for the best poster prize. I hope to come back next time to talk about even more innovations in crystal structure prediction!

031

Reposted by Jordan Dorrell

Juliana Morbec @jmmorbec.bsky.social · 16/09/2025

Debuting my bsky.social account with a photo from the first day of the 2025 Materials and Molecular Modelling (MMM) Hub conference at @keeleuniversity.bsky.social . Thrilled to be the local organiser for this exciting event.

061

Jordan Dorrell @stochasticchemist.bsky.social · 16/09/2025

A dreary Tuesday morning in Southampton but excited for the #MMMHub conference this afternoon! If you want to hear what I've been up to with @graemeday.bsky.social , I have a talk today at 15:40 and I'm poster number 7! #compchem

010

Reposted by Jordan Dorrell

Graeme Day @graemeday.bsky.social · 12/09/2025

Thank you to the organisers of the "from molecules to materials" meeting in Bologna for inviting me to give a keynote talk. Among other things, I spoke about @aichemyhub.bsky.social-funded large-scale crystal structure prediction and transferable ML potentials: doi.org/10.1039/D4FD... #compchemsky

Image of speaker at conference in front of projected slide showing computer simulation results on molecular crystals.

0144

Jordan Dorrell @stochasticchemist.bsky.social · 10/09/2025

Article doesn't mention anything about the UK being bad for research. Merck just complains that the UK doesn't pay enough for drugs and that Trump is pressuring them to invest in US facilities instead.

000

Jordan Dorrell @stochasticchemist.bsky.social · 10/09/2025

I don't think it will replace anything. Just how LDA, GGAs, and meta-GGAs can coexist, FFs, MLIPs, and DFT will coexist. MLIPs will just take over some of the tasks where FFs aren't accurate enough and DFT isn't fast enough.

131

Jordan Dorrell @stochasticchemist.bsky.social · 12/08/2025

Interested in CSP? I have a talk and a poster at the MMM Hub conference next month on a new method for improving the efficiency of structure prediction of crystals with more than 1 molecule in the asymmetric unit. See you there! thomasyoungcentre.org/event/mmm-hu... #compchem #csp #mmmhub

041

Reposted by Jordan Dorrell

Graeme Day @graemeday.bsky.social · 07/07/2025

Attending the @rsc.org International conference on materials chemistry #MC17, like organic materials and structure prediction? Then check these out from our team: #Chemsky #CompChemsky 1/4

164

Jordan Dorrell @stochasticchemist.bsky.social · 28/06/2025

The modern mix of GPUs and CPUs definitely complicates things. In CSP, people used to benchmark by number of force evaluations but that's not applicable to gen-ML. I wonder if there's any merit in comparing GPU and CPU methods by energy used? It's hardware specific, of course.

100

Jordan Dorrell @stochasticchemist.bsky.social · 27/06/2025

for the 8th test, it would be really nice to see compute time taken into account. Perhaps a finite amount of compute could be allocated for each group? This would help to separate the efficiency of the method from the amount of resources thrown at the problem. There's maybe better solutions though.

100

Jordan Dorrell @stochasticchemist.bsky.social · 27/06/2025

If your method can't find the experimental crystal(s), then what's the point? And if it's slower than other methods, why wouldn't I use the faster method? I think the 7th @ccdc.cam.ac.uk blind test made a really positive change in splitting up structure generation and structure ranking, but

100

Jordan Dorrell @stochasticchemist.bsky.social · 27/06/2025

It's interesting to see gen-ML models for CSP, but I feel that often there is insufficient evidence to show the efficacy of the method. For any CSP method, I think the two most important questions are: 1: Does it find the experimental crystal(s)? 2: Is it faster than other methods? #compchemsky

100

Jordan Dorrell @stochasticchemist.bsky.social · 24/06/2025

Can you do anything to cut the sourness or is this a final product?

100

Reposted by Jordan Dorrell

Chemical Science @chemicalscience.rsc.org · 18/06/2025

This week's #ChemSciPicks comes from @graemeday.bsky.social (University of Southampton), @ffmmgg.bsky.social‬, Chengxi Zhao‬, Xenphon Evangelopoulos, and @aicooper.bsky.social‬ (University of Liverpool). Read the full paper here: doi.org/10.1039/D5SC... #ChemSky

174

Jordan Dorrell @stochasticchemist.bsky.social · 17/06/2025

Someone was just telling me yesterday that I should think about cloud computing! I'd be interested to listen.

110

Reposted by Jordan Dorrell

Mihails Arhangelskis @marhangelskis.bsky.social · 17/06/2025

Join our group as a postdoc! Computational design of hypergolic MOFs through the use of machine learning and crystal structure prediction. If you are interested in ML for materials design, this could be for you: euraxess.ec.europa.eu/jobs/351559 #chempostdoc #compchem #CompChemJobs

euraxess.ec.europa.eu

Post-doc (adiunkt)

An announcement for postdoc position in the project entitled "Computational design of next generation rocket fuels based on hypergolic metal-organic frameworks" financed by National Science Centre is ...

043

Reposted by Jordan Dorrell

CCDC Cambridge @ccdc.cam.ac.uk · 10/06/2025

New, searchable fields have been added to existing CSD entries, adding scientific value and analytical potential. This enrichment allows researchers to interact with structures in entirely new ways. 👀 ccdc-info.com/3HotDJD #Crystallography #CompChemSky

Promotional graphic showing part of a word cloud of structural chemistry terms.

144

Jordan Dorrell @stochasticchemist.bsky.social · 10/06/2025

Great additions! Are there any plans to add G/Z''? I'd find it immensely helpful to be able to search the CSD by number of molecules in the asymmetric unit, rather than just number of formula units in the asymmetric unit.

300

Reposted by Jordan Dorrell

Derek Lowe @dereklowe.bsky.social · 28/05/2025

Materials science can be a great place to apply AI/ML techniques. But if you think that makes it less prone to hype and sloppy work, well. . . think again:

science.org

Whoa Now: Cautionary Tales from Materials Science

22110