As I've grown older, I've noticed an ever-increasing frequency of people brandishing tattoos. My daughters have been trying to convince me to join them and get matching tattoos of dancing Grateful Dead bears on our necks. After reading about the ancient Maya's tattooing practices, I might join them—a sort of life event mark.
Speaking of a life event, I am now all done with chemo. My body scans have not shown signs of cancer for about a year and a half now. I kept up my treatment as a precaution to make sure it had been eliminated. I will now be getting scans every 3 months to stay on top of it. I am grateful to the Oncology team in Hibbing. I have developed friendships with several over the last 4 years.
I wake up every day and think I'd better make the most of it because I might not wake up the next day. Cancer fundamentally changes your life, and in my case, it was a very positive change.
a) Stela 11 captives with tattoos or scarification at Yaxchilan, Mexico b) Captives on Stela 12 at Piedras Negras, Guatemala. Credit: C. Helmke, in Stemp et al. 2025
MAPS
The model is currently in the garage for maintenance. I installed a new positive definite scheme, but in doing so, I inadvertently created a memory leak. Additionally, some minor errors in the solver have been corrected. The model is now set up to handle SARS-CoV-2, measles, influenza A, and H5N1 (when it goes from human to human).
At the WHN, they are creating an interactive interface that will allow you to view New Daily Infections for the entire US, down to your location, spatially and graphically. Eventually, this interface will go global and include warning boxes around new outbreaks. The model forecast will be run 2-3 times a week.
Let us stay on modeling for a bit longer. A paper came out a couple of days from the Sato Lab’s Jumpei Ito as the lead author. The paper details the use of a protein language model to detail the viral fitness landscape.
Successively emerging SARS-CoV-2 variants lead to repeated epidemic surges through escalated fitness (i.e., relative effective reproduction number between variants). Modeling the genotype–fitness relationship enables us to pinpoint the mutations boosting viral fitness and flag high-risk variants immediately after their detection. Here, we present CoVFit, a protein language model adapted from ESM-2, designed to predict variant fitness based solely on spike protein sequences. CoVFit was trained on genotype–fitness data derived from viral genome surveillance and functional mutation assays related to immune evasion. CoVFit successively ranked the fitness of unknown future variants harboring nearly 15 mutations with informative accuracy. CoVFit identified 959 fitness elevation events throughout SARS-CoV-2 evolution until late 2023. Furthermore, we show that CoVFit is applicable for predicting viral evolution through single amino acid mutations. Our study gives insight into the SARS-CoV-2 fitness landscape and provides a tool for efficiently identifying SARS-CoV-2 variants with higher epidemic risk.
a Conceptual framework of CoVFit. CoVFit is a protein language model designed to predict the fitness (relative Re) of SARS-CoV-2 variants based on their S protein sequences. b Outline of the training process used to develop CoVFit model instances.
Now here is the kicker, the Spearman rank score was nearly 1! I flipped out when I saw this. The also applied it to individual countries and the results indicated a high scores.
To evaluate the prediction performance of the CoVFitNov23 model instances, we examined the prediction performance of respective model instances using the corresponding test datasets. We selected Spearman’s rank correlation score as the main metric because we believe it is necessary to assess the model’s ability to predict the ranking of relative fitness between variants in order to prioritize variants with high fitness that are likely to spread. The resulting prediction performance for fitness on data that doesn’t require extrapolation is notably high (Spearman’s correlation: 0.990) (Fig. 2a–c). Also, prediction of escape ability from neutralization by mAbs reaches moderately high performance (Spearman’s correlation for each epitope class: 0.578–0.814) (Figs. 2a and S3a). The model’s higher predictive performance was also supported in evaluations stratified by sampling country and mAb type for fitness and immune escape ability, respectively (Fig. S3b, c). Even when variants were stratified by country of circulation and emergence time (on a weekly basis), a robust correlation was observed between fitness and predicted fitness (Fig. S3d, e). This result suggests that CoVFit can distinguish the fitness of variants emerging during the same period with an effective level of accuracy. Together, we show that CoVFit has sufficient power to represent the fitness landscape as well as the effect of mutations on evasion from diverse types of mAbs.
a Spearman’s correlation scores for predicted relative fitness values and mAb neutralization escape scores. Scores from five cross-validation folds are shown as dots, with the mean represented by a bar and the standard deviation by an error bar. The correlation for mAbs was calculated in each epitope group. b Scatter plot for fitness prediction, aggregating results from five-fold cross-validation. Dot denotes the result of a certain viral genotype in a specific country. Dot is colored by the Nextclade clade. The relative fitness value was scaled so that the 0.1 percentile and 99.9 percentile points fall between 0 and 1. A dashed line with a slope 1 and intercept 0 is shown. c Scatter plot inherited from (b) but colored by the emergence date of each genotype. Source data are provided as a Source Data file.
I need to contain my enthusiasm because this model brings the possibility of long-term integration of MAPS by introducing the possibility of predicting future variants.
Once I figure out how to couple MAPS and CoVFit I will be able to run an ensemble of 20+ year integrations, similar to how a Global Climate Model (GCM) uses an ensemble of integrations to predict climate. MAPS+CoVFit could also be used to simulate the long-term effects of mitigation strategies. This type of system would be of great interest to many outside of the science community, such as the long-term cost to our health systems.
The CoVFit model is also useful in the short-term MAPS forecast system because it can estimate the relative growth of new emerging variants almost instantaneously. Currently, we have to wait for a large pool of sequences to be uploaded to GISAID, which is used to calculate the Immune Escape and ACE-2 binding affinity of the variant. These are the two quantities that enable MAPS to set about 30 variant-specific settings automatically.
I would like to thank the Sato Lab and Jumpei Ito’s team for their continued excellent work. Pssst, can you get me some data on NB.1.8.1?
SARS-CoV-2 Update
LP.8.1 continues to be the dominant variant, while XEC is in a free fall. Here is the CDC’s variant proportions analysis.
The last time I wrote this newsletter we were quite a bit different.
There are some in the know that have suggested that the CDC has been in error for some of the recent variants.
Here are the current nationwide wastewater derived New Daily Infections (NDIs).
Here is a snapshot of the current New Daily Infections across multiple states. The yellow areas are preliminary estimates. Overall, the picture is "good" compared to where we were.
Overall, the data indicates a general trend towards lower values. We have to remember to keep this in perspective. Nationwide, the NDIs are still about twice the values during the first significant wave. Is 500-600K NDIs our new “baseline”? It sure appears that way. Below is an analysis since the first of the year. So, let me ask: is ~200,000 deaths since the start of the year acceptable? Do you realize the costs of 5-25 million new long COVID patients? I'd better get MAPS+CoVFit to do some estimates over the long-term or decadal scale.
Get ready for this “lull” to change in the coming 2 months. There is a variant named NB.1.8.1 that has been taking off in different parts of the globe. The last I saw its relative growth advantage was 50% higher than LP.8.1. Here is good description from Xu Zou
Here is a little more on this variant.
To me, NB.1.8.1 deserves a new name. We are heading towards the train wreck we witnessed last summer, when we experienced the double peak of over 3 million new daily infections (NDIs). It is interesting to examine the plot below by state, which indicates that specific regions of the country drove the six local maxima. This affirms what I have been stating over the last 4 years. The pandemic is global, but the evolution of the virus is local. There is a concept referred to in the Sato Lab paper as the “immune landscape,” which I consider to be a combination of various regional factors, including masking, vaccine mitigation, and weather conditions.
I would like present a summary of a short study posted by tern (@1goodtern in X). I have saying for quite some time in this venue, and in interviews by The Matt McNeil show, that the continuing battering of our immune system will open the door for many other pathogens. The tern study clearly shows the dramatic effect of the pandemic on other pathogens.
A lot of people have noticed it. In the past few years, a bunch of different infections in England have started moving together. Spikes, dips, rising and falling in sync.I had the data handy for a few infections going back over the last 15 years, so I thought I'd run a pearson correlation on each of the pairs.
This is how much correlation there was among them *before* Covid arrived. This is the correlation during the period 1/1/2016 to 1/3/2020.
What's this saying and how does it work...
Scroll your finger down the left side to pick an infection... like Measles... and then scroll along to pick something to see whether it's spread correlates with that condition... like Hep E.
So before Covid arrived, their correlation was +0.07. Which means there's no significant matching pattern. It's coloured kind of light brown. Blue means it's perfectly out of sync... and red means it's perfectly in sync.
The darker blue, the more precisely this is up when that is down and vice versa...
And the darker red, the more precisely this is up when that is up, and this is down when that is down.
So things are a bit all over the place before Covid... Some things are completely out of sync. And some are completely in sync. And some things are not at all in sync.
And remember, these are *annual* rolling totals, so this isn't about time of year, it's about yearly trends.
How do things change when Covid arrives...
WOW! The mean correlation is now 0.79 vs 0.07. Folks, this is not sustainable in the long term, both emotionally and financially. If someone states that the pandemic is over, send them a link to this Substack. We are in dire need of educating and waking people up from their pandemic denial. Future generations are inheriting this mess.
On average, individuals have had ~4 infections since the start of the pandemic, and that value continues to accumulate and will likely accelerate. Think about it. When SARS-CoV-2 first entered the world stage, the asymptomatic rate was south of ~10%, now I am seeing values from different parts of the globe indicating north of 70%! People are walking around thinking they have a mild case of the flu when it is COVID. Many folks consider COVID to be like the flu. This lie needs to be called out. As I have stated, even a “mild” infection can cause severe damage to our organs. People don’t get that. And the damage is not a one-shot deal; it can be ongoing through viral persistence. Are you OK sending your children to school knowing that they will get at least 1 SARS-CoV-2 infection during the school year? Please read the following post from a brilliant scientist (she also left the country to pursue her work largely because of the current administration's cuts to science) and say that COVID is like the flu.
Time is wasting, get to work and educate your friends and family, a tipping point could be right around the corner.
I have been stressing to others lately that the current administration’s slashing of science funding, with a strong emphasis on health sciences and services, will take a LONG time to recover from. Check out this editorial I read yesterday that emphasizes just that point.
I ask again, are you comfortable sending your kids to school? Demand updates to your school's HVAC system, push for universal masking in schools, and demand HEPA filters in every classroom!
I will be back on a regular schedule now; my life has been rather hectic the last couple of months and includes dealing with bad case of the norovirus. I appreciate the many well wishes I have received.
And now for something completely different …
I will be collecting dancing bear designs for my first tattoo. Please send your ideas to me :-)
From ethnographic accounts and ancient Mayan imagery, we know that the Maya engaged in various forms of body modification, including piercing, body painting, scarification, and tattooing. However, no Mayan remains with tattoos or tattooing implements are known from the archaeological record.
Despite this, we know tattoos were found on a wide variety of individuals and differed depending on the individual receiving them, serving both as a source of pride and a form of punishment.
Dr. W. James Stemp elaborates, "Based on the Spanish ethnohistoric accounts and Classic Maya artwork, it appears both adult men and women could be tattooed. We have no evidence suggesting children were tattooed.
"Men and women seem to have been tattooed differently in terms of body parts, and this also seemed to vary somewhat by ethnic/language group. It seems that tattooing was done to mark transitions in a person's life or to mark important events, and that tattooing may have been both a sign of value, as well as a form of punishment, depending on the person and the situation.
Next up will be a weekend forecast with the latest CDC variants proportions analysis with some new goodies for you.
Here is a kick-ass at-home test, nearly equivalent to a PCR test.
SARS-CoV-2 is a vascular virus, not just a respiratory virus. To make it simpler for folks, call it a whole-body virus. No organ is spared. It infects our entire body and creates reservoirs throughout. It should be thought of as a Long COVID virus and will exploit pre-existing risk factors.
Check out this article. I love it. It’s an easy read but stimulates the brain in many ways.
Good news on cancer front. I hate the rest in your update. It’s never ending and the societal pressures for our fam with 3 teens is becoming untenable. I don’t know how to know from experts what this virus is capable of, seeing it first hand in our fam and ignore it so we can participate in life. Lastly, I’m in Duluth … if you figure out a solution to obtain Plus Life tests, will you message me.
Good news on cancer front. I hate the rest in your update. It’s never ending and the societal pressures for our fam with 3 teens is becoming untenable. I don’t know how to know from experts what this virus is capable of, seeing it first hand in our fam and ignore it so we can participate in life. Lastly, I’m in Duluth … if you figure out a solution to obtain Plus Life tests, will you message me.
...And then there's this...
https://www.thelancet.com/journals/eclinm/article/PIIS2589-5370(24)00228-1/fulltext