If you're interested in integrating theory and data to better understand how cancers evolve, register for the 2nd CANPOP meeting!
Featuring a keynote talk from Martin Taylor (@mstaylor.bsky.social) and selected short talks, what better way to spend an afternoon?
forms.gle/jSxB1TdPURRa...
If you’re interested in machine learning and mathematical modelling, we’re recruiting a PhD student to develop new methods for studying cancer evolution at
@imperialimmuno.bsky.social
www.findaphd.com/phds/project...
An incredible opportunity for early career researchers - Imperial's Faculty of Medicine is hiring 20 Assistant Professors!
We held the first #CANPOP symposium today (cancer population genetics club) @icr.ac.uk. 6 excellent & didactic talks, all applying mathematical #popgen theory to dissect somatic evolutionary dynamics: spanning the full range of pre-cancer, metastasis through to 6000 year old transmissible cancer...
Thanks Marc!
Studying cancer evolution needs multi-region or single cell seq for phylogenetics, right? Amazingly (I think!) we found single-sample bulk methylation suffices, via analysis of "fluctuating methylation". In @nature.com today led by brilliant @calumgabbutt.bsky.social www.nature.com/articles/s41...
This was a massively collaborative project and it was a pleasure to work with such amazing researchers, but particular thanks to Inaki Martin-Subero, Martí Duran-Ferrer @idibaps.bsky.social and @trevorgraham.bsky.social
Finally, fCpGs record clonal dynamics over time. In two patients with Richter transformation (RT), the emergence of an altered phenotype with dismal outcomes, we inferred that the RT clone diverged from the non-RT lineage over 30 years prior to its clinical manifestation! (6/7)
In chronic lymphocytic leukaemia (CLL) the high-risk U-CLL subtype had much higher growth rates than the low-risk M-CLL subtype. Stratifying by growth rate within these groups was highly prognostic of the time to first treatment. (5/7)
Across 2000 lymphoid cancer samples, we found staggering heterogeneity between different diseases and molecular subtypes! Paediatric (ALL) grew much more rapidly than adult cancers, and the aggressive 11q23/MLL subtype grew even faster than the other subtypes. (4/7)
We designed a computational method, EVOFLUx, to infer the early evolutionary history of a cancer from these data. For instance: when did the most recent common ancestor emerge and how rapidly was the cancer growing at that time? Is the cancer undergoing a subclonal expansion? (3/7)
We used DNA methylation “evolving barcodes” to record the lineage of cells, which we termed fCpGs. These could be measured using low-cost, bulk methylation arrays. The clonal dynamics of cells are recorded in the patterns of these fCpGs (2/7)
Cancer is an evolutionary disease, but does knowing a cancer’s evolutionary past help predict its future? Out today in @nature, we learnt the evolution of 2000 lymphoid cancers and found it was highly correlated with clinical outcomes! (1/7)
rdcu.be/eFrrc
🚀 AI in Science Fellowships – Applications Now Open!
The I-X Centre for AI in Science is recruiting up to 19 fellows to join their prestigious programme and accelerate artificial intelligence research in Engineering, Natural and Mathematical Sciences.
1/5
Imperial's Schmidt AI in Science fellowships are now open! As a current fellow, I can highly recommend applying for these prestigious and enriching awards. No prior AI experience needed, just a scientific problem suited to AI. For cancer researchers there are dedicated joint positions with the ICR.
Schmidt Fellowships for #AI in #cancer research 2025 round is now open. These are pathway to independence fellowships between @icr.ac.uk & @imperialcollegeldn.bsky.social that support fellows to be competitive for faculty jobs at the end of the 2 year funding. www.imperial.ac.uk/jobs/search-...
Darwin's famous sketch of a rooted tree, representing speciation
New preprint! Let me tell you a story about trees, caterpillars, brooms, entropy, and getting scooped by 50 years.
Rooted trees of all shapes and sizes crop up all over biology, computing and elsewhere. How can we best compare the shapes of these myriad trees? 1/14
arxiv.org/abs/2507.08615
Interested in population genetics of cancer?
If so, have a look at our manuscript accepted by @genetics-gsa.bsky.social, in which we provide a theoretical assessment of the genetic makeup of cancers before and after treatment!
Calling data scientists: we have opening in our Data Science core @icr.ac.uk to lead work around cancer spatial biology and single cell analysis. These are staff scientist type positions, ideal for someone who wants to help drive computational research. Apply here: jobs.icr.ac.uk/vacancies/12...
If you're a PhD student interested in using maths to understand cancer, this HCEMM summer school in Szeged, Hungary, is an excellent opportunity.
Registration is free and local costs are covered, so apply before 15th April.
This week-long summer school looks set to be an excellent opportunity for to learn about cancer evolution!
Bursaries are available for those who require financial assistance, deadline 1st April.
Darryl never even made it to Twitter, so Bluesky may be a stretch! Really glad you liked the pre-print, if you've got any questions or would like to chat, please feel free to email me at calum.gabbutt@icr.ac.uk
And of course, thanks to all our institutes: the ICR, BCI, IDIBAPS, ASU, CIBERONC, Hospital Clínic de Barcelona, Uppsala University, USZ, Universitat de Barcelona, UCL, USC and ICREA. (9/9)
A big thanks to Martí Duran Ferrer, Heather Grant, Diego Mallo, Ferran Nadeu, Jacob Househam, Neus Villamor, Olga Krali, Jessica Nordlund, Thorsten Zenz, Elias Campo, Armando Lopez-Guillermo, Jude Fitzgibbon, Chris P Barnes, Darryl Shibata, José I Martin-Subero and @trevorgraham.bsky.social! (8/9)
To conclude, we present a cheap and general new technique to measure evolution in cancer from single-timepoint, bulk samples. This links basic cancer evolution research directly to translational medicine and may allow us to focus treatment on just those who need it. (7/9)
The inferred growth rate of the typically more aggressive U-CLL cancers is greater than that of the often indolent M-CLL cancers.
A Kaplan Meier curve showing the inferred growth rate stratifies time to first treatment in CLL patients, controlling for IGHV mutational status.
In B-ALL, MCL and CLL, different clinical subtypes had markedly different growth rates and effective pop sizes. In CLL, the growth rate was highly prognostic of time to first treatment, whilst the pop size was a better predictor of over survival. (6/9)
A: Timeline of two CLL patients with samples collected longitudinally, annotated with treatment received. Circles represent methylation array (blue: 450K, orange EPIC), squares represent whole genome sequencing, treatment is represented with a green vertical line and Richter transformation (RT) is represented with a vertical pink line. B: (left) The reconstructed phylogenies of the relationship between samples, annotated with the clinical classification of each sample. The black triangles represent the time that occurred since the most recent common ancestor, taken as the posterior median of T-τ from the single-sample EVOFLUx inferences. (right) A heatmap representing the 978 fCpG loci, with the colour a representing the fraction methylated (0% blue, 100% red).
We also able to infer the phylogenetic relationship between longitudinal samples. In CLL, some patients undergo Richter transformation (RT), the emergence of an aggressive phenotype – this lineage diverged >30 years prior to clinical detection in 2 CLL patients. (5/9)
A scatterplot showing the inferred evolutionary parameters of 1,976 lymphoid malignancies.
Boxplots comparing the distribution of subclonal weightings inferred by EVOFLUx in samples called as neutral vs under subclonal selection via the WES data.
Applying EVOFLUx to quantify the evolution of 1,976 lymphoid malignancies, we found widespread heterogeneity between and within cancer types. Our subclonal inference was validated with matched deep whole exome sequencing. (4/9)
An example of our model's fit (posterior predictive) to real patient fCpG methylation data.
A pairs plot of the posterior of the inference run on simulated data, with the ground truth values used to generate the simulated data highlighted in red.
We developed a new computational method (EVOFLUx) to infer a cancer’s evolutionary history from methylation data. We can learn how quickly a cancer is growing, when its most recent common ancestor existed, its effective pop size and the presence of subclonal expansions. (3/9)
Illustration of how fCpGs enable tracking of lineage dynamics, describing three limiting cases of a population evolutionary structure: (top) polyclonal, in which the most recent common ancestor (MRCA) exists far in the past; (middle) clonal, in which the MRCA is very recent; (bottom) ongoing evolution following a population bottleneck.
Evolution is a dynamic process, but often clinicians only have a single snapshot of a tumour. In our paper last year, we discovered fluctuating methylation as a natural “evolving barcode”, which encodes the evolutionary history of a set of cells. (2/9)
doi.org/10.1038/s415...