Chris Wymant

We're hiring a modelling postdoc at PSI Oxford for two exciting projects: 1) modelling the early immune responses to Nipah vaccination, and 2) joining the PRESTO team working on immunobridging in vaccine evaluation studies.

tinyurl.com/5abbxrjh

Get in touch for more info! Deadline 4th August.

Please circulate to anyone interested: Senior Statistician role at UK Health Security Agency national Real-time Syndromic Surveillance Team. This would be a great job for any modellers or statisticians with an interest in surveillance.

www.healthjobsuk.com/job/UK/Londo...

Infectious Disease Modelling starter pack update! Many more modellers have joined Bluesky recently so this update is overdue. Pls keep on sending suggestions! (bio should contain experience relevant for this pack)
IDModelling pack 1: go.bsky.app/86Ao1a5
IDModelling pack 2 : go.bsky.app/2oBB7KX

John Drake shaking hands

Congrats to @jdrakephd.bsky.social on the important achievement of winning our* infectious disease pub quiz. Honorable mentions to Corin Yeats, Luca Ferretti, @christophraser.bsky.social and @aliciagill.bsky.social
* the Pandemic Sciences Institute; Data, Epidemiology and Analytics section

Testing samtools’ base alignment quality for HIV Testing the appropriateness of samtools’ base alignment quality (BAQ) calculation for HIV Recap: BAQ calculates a per-base probability of being misaligned, converts that to a quality score, then repl...

Virus genomics peeps: if you use samtools pileup to get base counts from files of mapped reads, you might want to switch off its calculation & use of 'base alignment quality'. I decided this was best for HIV, documented here (in 2019, better late than never) docs.google.com/document/d/1...

It's official: the #PandemicAccord is officially adopted by the World Health Assembly!

My warmest congratulations to @who.int Member States for their commitment to keeping their people and the world safer.

What a moment in global health history. Together!

First publication 10. How to be polite and show respect * In the old days of feudal Japan, a samurai warrior would shout You are trying to view Flash content, but you have no Flash plugin installed.

I had no idea samurai in feudal Japan were that dedicated to troubleshooting Flash

Come work with us! We're looking of postdoc(s) with experience in bacterial genomics and interest in mathematical modelling to work on bacterial ecology and evolution in the context of public health. tinyurl.com/n34vzszu

An astronomy professor colleague of mine once relayed trying to explain to his students why it was important that they actually write their class reports themselves. “The point is not to teach ME about neutron stars,” he said.

Academic paper/talk titles since time immemorial:
SUCCESSFUL PREDICTION OF THINGS THAT HAVE ALREADY HAPPENED

How should the advancement of large language models affect the practice of science? | PNAS Large language models (LLMs) are being increasingly incorporated into scientific workflows. However, we have yet to fully grasp the implications of...

An LLM "creates textual claims, and then predicts the citations that might be associated with similar text. Obviously, this practice violates all norms of scholarly citation.

At best, LLMs gesticulate toward the shoulders of giants."

Bender, West, and I contributed to this pro/con piece in PNAS.

Two postdoc positions to work on virus epi & evolution in response to vaccination, with both theoretical models + data analysis. Paris/Montpellier. With Sylvain Gandon, Sébastien Lion, François Blanquart, Katrina Lythgoe, & Troy Day
emploi.cnrs.fr/Offres/CDD/U...
emploi.cnrs.fr/Offres/CDD/U...

Does your team deposit and access pathogen genomic data via data-sharing platforms for public health use? Participate in WHO's online consultation on the critical features of pathogen genomic data-sharing platforms to meet public health needs.

All info here: www.who.int/news-room/ar...

Please spread the word - a funded (home fees) DPhil (PhD) studentship available in @oxfordstatistics.bsky.social

Social optimisation of public-facing digital tools for health protection and trial frameworks for non-pharmaceutical interventions
www.stats.ox.ac.uk/research-stu...

Unpacking sources of transmission in HIV prevention trials with deep-sequence pathogen data – BCPP/ Ya Tsie study To develop effective HIV prevention strategies that can guide public health policy it is important to identify the main sources of infection in HIV prevention studies. Accordingly, we devised a statis...

Thrilled that this work by Lerato Magosi with many colleagues is now out as a preprint www.medrxiv.org/content/10.1....
It suggests results from treatment-as-prevention HIV cluster-randomized trials are lower than many expected due to significant "spillover" of infections from untreated communities

Excited to share two new preprints from our team this week, both focusing on the shifting #HIV epidemic among adolescents and young adults (AYA) in #Uganda. Good news is the shift is happening in the right direction, as a result of both shifting behaviors and increasing coverage of interventions.

Viral and immune dynamics of genital human papillomavirus infections in young women with high temporal resolution Human papillomavirus (HPV) infections drive one in twenty new cancer cases, with a particular burden on women. This analysis of longitudinal clinical data using kinetics models reveals that HPV infect...

New Alizon-lab HPV paper in @plosbiology.org: 2 years of follow-up of young women in France, 76 infected with HPV. Viral load increases rapidly on infection (<2 months) then plateaus for 13-20 months in those clearing. Dynamics of immune cells, cytokines & Abs too
journals.plos.org/plosbiology/...

Top panels: graphs showing increases in spillover events, extinction rates, and temperature anomalies over the last few centuries. Bottom panel: a map of 10 pandemics since the year 1900. Four were linked to agriculture, two to wildlife use, and one to climate change.

🚨😷🧪 NEW: A growing body of evidence shows that pandemics, biodiversity loss, and climate change are part of a broader polycrisis - but there are no simple solutions. A sweeping overview of "Pathogens and planetary change" for the first issue of @natrevbiodiv.bsky.social, out now 🔓 rdcu.be/d6lHl

Google search for "caffeine per day" AI Overview gives potentially deadly advice, suggesting that 10 energy drinks per day is 400mg caffeine. It's actually around 2000. "The recommended daily caffeine intake for healthy adults is up to 400 milligrams (mg). This is equivalent to about four cups of brewed coffee or 10 cans of energy drinks. However, individual caffeine tolerance varies, and some people may experience negative effects at lower doses. It is important to listen to your body and adjust your caffeine intake accordingly. Pregnant women, children, and people with certain medical conditions may need to limit their caffeine intake further. Consult with a healthcare professional for personalized advice."

How is this production technology?

Here's Google Gemini potentially getting someone killed. Ten energy drinks is typically 1600-3000 mg of caffeine.

As with most epi forecasts, ranges often come from differences in model structure/data not sampling.

The same is true for elections, climate change, and many other domains.

We need more focus in our intro stats classes on these epistemic sources of noise, so people can better consume forecasts.

Job Opportunity at LSHTM: Professor (and Head of Department) The London School of Hygiene & Tropical Medicine (LSHTM) is one of the world’s leading public health universities. Our mission is to improve health and health equity in the UK and worldwide; working i...

Job opening for full professor and head of dept in infectious disease epidemiology and dynamics at @lshtm.bsky.social. Deadline 31st Jan: jobs.lshtm.ac.uk/vacancy.aspx...

(Redraw if all(TRUE) || all(!TRUE). Bluesky not the best place for while loops 8-)

Can use e.g. rdrr.io/snippets/ on your phone
Tell them "what does everyone in one group have in common, that they don't have in common with the other group?"
Num partitions = sum(choose(N, 1:floor(N/2))))
Larger N is also harder for knowing everyone in the group. N~10 is good.

Festive family and friends fun that nerds can help administer: randomly partition a (WhatsApp) group in two, and the challenge is to think up a criterion that would result in this partition
library(tidyverse)
tibble(person = c("Joe", ...)) %>%
mutate(group = rbernoulli(nrow(.))) %>%
arrange(group)

It's a rare day when Adam Kucharski and my daily chocolate are equally informative

The Institut Pasteur is recruiting group leaders!
Pasteur is an absolute incredible environment to start your lab. Surrounded by so many great scientists, in the beautiful city of Paris.
Come eat 🥐 and conduct awesome science.
Contact me if you want info !

research.pasteur.fr/en/call/call...

New @cgdev.org policy paper:

*MALARIA VACCINES* are a huge scientific breakthrough. But the world is dragging its feet on rolling them out.

We examine why, and how to go faster.

🧵

Image shows how the probability of reported infection with SARS-CoV-2 among people notified by the NHS COVID-19 App increases with the risk score calculated by the app.

👋 Hello World! Some of our papers this year ICYMI
Ferretti & Wymant et al, Nature: using 7 million digitally recorded COVID exposures, we found app-based contact tracing successfully measured the actual risk of transmission and we decomposed contributors to risk 045.medsci.ox.ac.uk/risk_measure...

I try not to be fussy about not being cited but making the exact same point *with the exact same title* fifteen months later in a sister venue?

That feels like some—what’s the right word?—it’s on the tip of my tongue…

undark.org/2023/04/06/c...

www.scientificamerican.com/article/chat...

When giving a talk, never say "as you all know". Introductory remarks are often familiar to many, there's no need to point this out. And you don't know the full background of every individual in the audience, so don't risk making anyone feel dumb.
("As we all heard in the previous talk" is OK.)