Carl Zimmer writes about AlphaGenome

We may not know a lot about how artificial intelligence (AI) algorithms work but the one thing we do know is that they are only as good as their databases. If you ask an AI program to tell you when Charles Darwin was born then chances are good it's going to give you the correct answer because that information is in Wikipedia and lots of other reliable online sources.

However, if you ask it to tell you how many genes are in the human genome it will not give you the correct answer. The correct answer is that we don't know for sure because it depends on how you define a gene and how many non-coding genes there are using various definitions. That's not the answer you will get. (I personally believe that there are only about 1000 non-coding genes but I don't expect a good "intelligence" program to favor my view over others. I DO expect it to not favor other opinions over mine.)

I just asked ChatGPT and it told me that there are tens of thousands of non-coding genes based on the Human Genome Project plus GENCODE and Ensemble annotations. This is correct ... and misleading. It's giving the best answer it can based on the databases it searches. However, many of us are skeptical of the GENCODE and Ensemble annotations and for good reason. They tend to err on the side of inclusion in order to avoid false negatives. In other words, they don't want to risk ignoring a real biologically relevant feature for lack of evidence so they deliberately risk including a lot of false positives. This is why those databases include a lot of questionable features such as non-coding genes, multiple transcription start sites, multiple splice variants, and tons of potential regulatory elements.

Along comes AlphaGenome. It's an AI program designed to scan those GENCODE and Ensemble databases to identify important features that might play a role in genetic diseases. What could possibly go wrong? [How intelligent is artificial intelligence?] [Will AlphaGenome from Google DeepMind help us understand the human genome?]

The average science writer jumped all over the original announcement of AlphGenome to let us all know that artificial intelligence was going to solve the problem of the mysterious genome. Apparently the complexity of the human genome has astonished scientists ever since the first human genome sequence was published 25 years ago.¹ The typical article on AlphaGenome fits nicely into the common theme that AI is soon going to rule the world.

That's why I was excited to pick up my copy of the New York Times yesterday and see that Carl Zimmer had written about AlphaGenome. Finally, an intelligent, highly respected, science writer was going to give us the truth. Here's the article that I saw in my version of the paper. (It was originally published several weeks ago on January 28, 2026.)

What a disappointment! Zimmer goes with the hype about AlphaGenome and repeats some of the tropes that he has avoided in the past. For example, he writes about how alternative splicing can create hundreds of different proteins from a single gene and how regulatory sequences can lie thousands or million of base pairs away from a gene. (There's no question that this is true for a small number of transcription factor binding sites but the vast majority are close to the promoter.)

Zimmer gives an example showing that AlphaGenome identified a regulatory sequence for a gene called TAL1, implying that the program will help decipher the rest of the genome. The general tone of the newspaper article is that AlphaGenome will be of great help to scientists who want to understand the human genome.

I checked the online version of Carl Zimmer's article in order to prepare for this blog post. I was surprised to see that there were lots of things in the online version that weren't in the newspaper article. For example, Zimmer quotes my colleague Alex Palazzo saying that everybody uses AlphaFold to study proteins then later on in the article Zimmer notes that, "But the more scientists studied the human genome, the more complicated and messy it turned out to be." The newspaper article left out the words "and messy" and that's significant because junk DNA supporters like Alex Palazzo often refer to the human genome as "messy" and full of junk DNA and that's a very different perspective than opponents of junk DNA who emphasize things like "complicated" and "mysterious."²

Zimmer has an even more revealing section that's in the online version but not the newspaper version.

Peter Koo, a computational biologist at Cold Spring Harbor Laboratory in New York who was not involved in the project, said that AlphaGenome represented an important step forward in applying artificial intelligence to the genome. “It’s an engineering marvel,” he said.

But Dr. Koo and other outside experts cautioned that it represented just one step on a long road ahead. “This is not AlphaFold, and it’s not going to win the Nobel Prize,” said Mark Gerstein, a computational biologist at Yale.

AlphaGenome will be useful. Dr. Gerstein said that he would probably add it to his toolbox for exploring DNA, and others expect to follow suit. But not all scientists trust A.I. programs like AlphaGenome to help them understand the genome.

“I see no value in them at all right now,” said Steven Salzberg, a computational biologist at Johns Hopkins University. “I think there are a lot of smart people wasting their time.”

The end of the online article is quite different from the final paragraphs of the newspaper article. In the newspaper article, Zimmer describes the TAL1 result then ends it with the paragraph starting with "In reality." I've highlighted that paragraph in the quotations below from the online version.

The AlphaGenome researchers shared their TAL1 predictions with Dr. Marc Mansour, a hematologist at University College London who spent years uncovering the leukemia-driving mutations with lab experiments.

“It was quite mind-blowing,” Dr. Mansour said. “It really showed how powerful this is.”

But, Dr. Mansour noted, AlphaGenome’s predictive powers fade the farther its gaze strays from a particular gene. He is now using AlphaGenome in his cancer research but does not blindly accept its results.

“These prediction tools are still prediction tools,” he said. “We still need to go to the lab.”

Dr. Salzberg of Johns Hopkins is less sanguine about AlphaGenome, in part because he thinks its creators put too much trust in the data they trained it on. Scientists who study splice sites don’t agree on which sites are real and which are genetic mirages. As a result, they have created databases that contain different catalogs of splice sites.

“The community has been working for 25 years to try to figure out what are all the splice sites in the human genome, and we’re still not really there,” Dr. Salzberg said. “We don’t have an agreed-upon gold-standard set.”

Dr. Pollard also cautioned that AlphaGenome was a long way from being a tool that doctors could use to scan the genomes of patients for threats to their health. It predicts only the effects of a single mutation on one standard human genome.

In reality, any two people have millions of genetic differences in their DNA. Assessing the effects of all those variations throughout a patient’s body remains far beyond AlphaGenome’s industrial-strength power.

“It is a much, much harder problem — and yet that’s the problem we need to solve if we want to use a model like this for health care,” Dr. Pollard said.

The net effect of these differences is to transform the article from one that promotes AlphaGenome in the newspaper version to one that's far more skeptical in the online version. I believe that the online version is far more accurate and reflects the high standard that I expect from Carl Zimmer. I'm assuming that the newspaper article was edited for the New York Times supplement that I read and I'm assuming that Zimmer did not approve of that edit.

Note: The cartoon was generated by ChatGPT in response to the request, "draw a cartoon illustrating GIGO - garbage in garbage out."

Note: The photo is from 10 years ago when Carl was in Toronto working on his junk DNA article for The New York Times [Is Most of Our DNA Garbage?]. That's Alex Palazzo on the left, then me, Ryan Gregory, and Carl Zimmer on the right.

---

1. Most knowledgeable scientists were not astonished to learn that 90% of our genome really is junk and there are fewer than 30,000 genes.

2. See the last chapter of my book: "Chapter 11: Zen and the Art of Coping with a Sloppy Genome."

Protein concentration in bacteria is regulated primarily at the level of transcription initiation

The amount of a given protein in Escherichia coli depends on a number of factors such as the amount of mRNA and the rate of translation. The standard model of regulation is based on decades of study of individual genes and it reveals that the amount of protein is mostly dependent on the amount of mRNA that was translated. This, in turn, indicates that most regulation occurs at the level of transcription initiation.

It's now possible to look simultaneously at the characteristics of large numbers of protein-coding genes to see whether this generality holds. That's what Balakrishan et al. (2022) reported in a Science paper a few years ago. They looked at the characteristics of 1900 protein-coding genes in E. coli to see how protein concentration was regulated.

Paralogs and LUCA

We're interested in the last universal common ancestor of all life (LUCA). In theory, this is a species that gave rise to both Bacteria and Archaea. The general assumption is that this is a single species with a well-defined set of genes that can now be found in all, or almost all, living species.

There are some scientists who question that assumption because they see massive transfers of genes between "species" during the early history of life. This gives rise to a web of life and not a well-defined tree. [The Three Domain Hypothesis: RIP] [The Web of Life] If that model is correct, then the ancestor of all living species could be a group of species that contributed different genes to a pool of organisms that lived billions of years ago. Early Bacterial and Archaeal ancestors could have independently acquired some genes by horizontal gene transfer.

How intelligent is artificial intelligence?

Over the past few years I've been assessing AI algorithms to see if they can answer difficult questions about junk DNA, alternative splicing, evolution, epigenetics and a number of other topics. As a general rule, these AI algorithms are good at searching the internet and returning a consensus view of what's out there. Unfortunately, the popular view on some of these topics is wrong and most AI algorithms are incapable of sorting the wheat from the chaff.

In most cases, they aren't even capable of recognizing that there's a controversy and that their preferred answer might not be correct. They are quite capable of getting their answer from known kooks and unreliable, non-scientific, websites, [The scary future of AI is revealed by how it deals with junk DNA].

Others have now recognized that there's a problem with AI so they devised a set of expert questions that have definitive, correct, answers but the answers cannot be retrieved by simple internet searches. The idea is to test whether AI algorithms are actually intelligent or just very fast search engines that can summarize the data they retrieve and create an intelligent-sounding output.

Genetics and the human life span

It seems reasonable to assume that there's a genetic component to aging and the human life span. For example, it's clear that if you inherit bad genes (alleles) from your parents then your chances of living a long life will be diminished. It's also clear that a lot of deaths (short life span) are not due to alleles you inherit from your parents but to extrinsic factors such as accident, war or disease. It's true that some diseases, such as cancer, have a heritable component but even people with "good genes" can die of cancer.

What's not clear is how much of the underlying, intrinsic, component of life span is due to alleles you inherit from your parents. If you look at the average life expectancy of men and women in different countries you can see that the average life expectancy of an American is about 80 years but people in Japan can expect to live five years longer. Is that difference due to genetics, or better health care, or something else?

Is epigenetics one of the best ideas of the 21st century?

New Scientist has produced a special issue on the best ideas of the 21st century. Here's the complete list.

NCSE and the Nature of Science

The National Center for Science Education (NCSE, United States) is an organisaton devoted to defending evolution and climate change. It has a long and admirable history of defending evolution from creationist attacks and of keeping creationism out of American public schools. Part of that defense involves keeping religious teachers and scientists as allies by suppporting the view that there is no necessary conflict between religion and science. NCSE has been a strong supporter of methodological naturalism—the view that science is restricted to investigating the natural world. I'll refer to this view as accommodationism.

Can NATO defend itself without the United States?

American Secretary of State Marco Rubio thinks that the United States of America has been subsidizing the defense of Europe and it's time for European nations (and Canada) to start paying more to defend Europe against Russia. According to Rubio, the other NATO countries have been taking advantage of the United States by cutting back on defense spending and using the money to fund social programs. (Gasp!)

Yeah, but our like-minded partners have to have capability and that's been part of the problem is the erosion in European defense capabilities because they've taken vast amount of the monies that these are rich countries and because of the NATO umbrella they it gave them the flexibility to spend a tremendous amount of their revenues on social programs and not on defense. Now maybe that trend line will begin to change.

First, let's be very clear about the huge US military budget. Only a fraction of that money is being used in the direct defense of Europe from Russian aggression. Canada, Poland, and France do not need to spend billions on B2 bombers that spend most of their time sitting on airfields and only occasionally are used to bomb targets in Iran. They do not need to spend billions on nuclear powered aircraft carriers positioned in the Caribbean or the South China Sea. They do not need to pay for troops in Iraq and Syria. They do not need to give Israel almost four billion dollars a year in military aid. And they do not need to spend huge amounts of money to fund a National Guard that can be used to police their cities.

Second, the Russian invasion of Ukraine has shown us that the threat from Russia is not that serious. Russia can't even conquer a medium-sized country right on its border. (Although it can do a lot of serious damage.) The idea that Russia poses a serious military threat to the main NATO countries is ludicrous.

Third, for all its posturing about defending Europe, the United States has failed to provide Ukraine with the supplies and equipment needed to prevent a gradual Russian takeover of Ukrainian territory. I guess the Americans decided to use the money to fund social programs instead.

Fourth, Canada and the European NATO countries are pretty much done with the constant bullying and insults from American leaders. We may be on the verge of finding out whether the United States needs allies.

---

American Secretary of State Marco Rubio explains the Venezuelan protection racket

I was watching Marco Rubio testify before a Congressional committee this morning. I'm interested because I live in Canada and we are one of the countries in the Western Hemisphere that are now supposed to be under the thumb of the United States.

So I paid attention when Rubio explained how the protection racket works with Venezuela because Canada will soon be next on the list. I'll quote directly form his testimony below but here's how I think it works.

America is using its military power to blockade Venezuelan ports and prevent any oil from being shipped except under special circumstances. The sale of oil to foreign countries must be approved by the United States and the United States will determine the price—usually the current market price. The payment won't go to Venezuela, instead it will be deposited into special accounts, for example in Qatar, that will be controlled by members of the Trump administration.

Venezuela can apply to use this money but only for the benefit of the Venezuelan people. For example, Venezuela could use it to fund the police or purchase medicine and equipment from the United States. They can also use it to buy light crude oil that can be used to dilute their heavy crude. They used to buy this from Russia but now they are buying it all from the United States.

Members of the Trump administration will decide which projects to fund. Unused money from the sale of Venezuelan oil might become the property of the USA (or its leaders).

Marc Rubio forgot to mention the legal and ethical justification of such a scheme but I'm sure he'll explain it later on. Other countries in the Western Hemisphere are interested.

From The United States Department of State.

And one of the tools that’s available to us is the fact that we have sanctions on oil. There is oil that is sanctioned that cannot move from Venezuela because of our quarantine. And so what we did is we entered into an arrangement with them, and the arrangement is this. On the oil that is sanctioned and quarantined, we will allow you to move it to market. We will allow you to move it to market at market prices – not at the discount China was getting. In return, the funds from that will be deposited into an account that we will have oversight over, and you will spend that money for the benefit of the Venezuelan people.

Why was that important? Venezuela was running out of storage capacity, okay. They were producing oil. They were drilling oil. They had nowhere to put it. They had nowhere to move it. And they were facing a fiscal crunch; they needed money in the immediacy to fund the police officer, the sanitation workers, the daily operations of government.

And so we’ve been able to create a short-term mechanism. This is not going to be the permanent mechanism, but this is a short-term mechanism in which the needs of the Venezuelan people can be met through a process that we’ve created, where they will submit every month a budget of this is what we need funded. We will provide for them at the front end what that money cannot be used for. And they have been very cooperative in this regard. In fact, they have pledged to use a substantial amount of those funds to purchase medicine and equipment directly from the United States. In fact, one of the things they need is diluent*, or diluent* depending how you want to pronounce it. And that basically is the light crude that you need to mix with their heavy crude in order for the oil to be able to be mixed and moved. They’re getting – they used to get 100 percent of that from Russia. They are now getting 100 percent of that from the United States.

So we’re using that short-term mechanism both to stabilize the country but also to make sure that the oil proceeds that are currently being generated through the licenses we’ll now begin to issue on the sanctioned oil goes to the benefit of the Venezuelan people, not to fund the system that existed in the past.

---

The Third Way Evolution Conference

The Third Way of Evolution is a strange organization composed of mavericks who think they're not getting enough attention. Here's how they describe their movement.

The vast majority of people believe that there are only two alternative ways to explain the origins of biological diversity. One way is Creationism that depends upon intervention by a divine Creator. That is clearly unscientific because it brings an arbitrary supernatural force into the evolution process. The commonly accepted alternative is Neo-Darwinism, which is clearly naturalistic science but ignores much contemporary molecular evidence and invokes a set of unsupported assumptions about the accidental nature of hereditary variation. Neo-Darwinism ignores important rapid evolutionary processes such as symbiogenesis, horizontal DNA transfer, action of mobile DNA and epigenetic modifications. Moreover, some Neo-Darwinists have elevated Natural Selection into a unique creative force that solves all the difficult evolutionary problems without a real empirical basis. Many scientists today see the need for a deeper and more complete exploration of all aspects of the evolutionary process.

Why it's necessary to be intelligent in order to explain function in the human genome

I follow the posts of Intelligent Design Creationists in order to see whether they have finally begun to understand science. Imagine my excitement when this post appeared on the Science and Culture website!

Why Intelligence Is Necessary to Explain Nature’s Functional Information

Alas, my excitement was short-lived. The post is actually about a podcast interview with William Dembski and not somebody intelligent.

• On the Meaning of the Word "Function"
• The Function Wars: Part I
• The Function Wars: Part II
• The Function Wars: Part III
• The Function Wars: Part IV
• Restarting the function wars (The Function Wars Part V)
• The Function Wars Part VI: The problem with selected effect function
• The Function Wars Part VII: Function monism vs function pluralism
• The Function Wars Part VIII: Selected effect function and de novo genes
• The Function Wars Part IX: Stefan Linquist on Causal Role vs Selected Effect
• The Function Wars Part X: "Spam DNA"?
• The Function Wars Part XI: Stefan Linquist responds to my critique
• The Function Wars Part XII: Revising history and defending ENCODE
• The Function Wars Part XIII: Ford Doolittle writes about transposons and levels of selection
• The function wars are over

---

These AI predictions are becoming ridiculous!

The first issue of Nature in 2026 has an article by science writer David Adam.

The Science of 2050
Nature explores the future breakthroughs that could shape our world.

The online version has a different title and subtitle but the text is the same. It begins with a quote from "futurologist" Nick Bostrum.

“There’s a good likelihood that by 2050, all scientific research will be done by superintelligent AI rather than human researchers. Some humans might do science as a hobby, but they wouldn’t be making any useful contributions.”

There's no attempt in the article to apply critical thinking to such a ridiculous prediction and the author doesn't consider the implications. If Bostrum (whoever that is) is right then that's the end of graduate studies and after 2050 nobody will be getting a Ph.D. in physics, biology, geology, or chemistry.

I hope I live long enough to see AI collecting and analyzing fossils in Greenland or studying volcanoes in Hawaii. Maybe I'll still be around when AI figures out how memories are stored or which transcription factor binding sites are functional in the human genome. And if I'm very, very lucky I'll see live to see all of my colleagues in the Department of Biochemistry abandon their labs and take up some scientific hobby like alchemy or intelligent design.

David Adam and the editors of Nature should be ashamed of themselves for publishing such nonsense.

---

Teaching the nature of science vs the scientific method

There's been a lot of talk about how to teach science literacy. The discussion in the USA centers around STEM (science, engineering, technology, mathematics) and this acronym has also spread to other countries. It's an unfortunate development since there's a big difference between teaching science and teaching those other three topics.

Most studies suggest that we focus on teaching The Nature of Science (NOS). There's no definition of this topic that everyone agrees to but the essence is that students need to understand how our society generates knowledge. In the context of the natural sciences, this means understanding the process of discovery. There's general agreement that what this means is critical thinking that's evidence-based. It's another way of saying that we need to teach critical thinking and the importance of using evidence to back up and test your claims of knowledge. "Appreciating the scientific process can be even more important than knowing scientific facts. People often encounter claims that something is scientifically known. If they understand how science generates and assesses evidence bearing on these claims, they possess analytical methods and critical thinking skills that are relevant to a wide variety of facts and concepts and can be used in a wide variety of contexts.”

National Science Foundation, Science and Technology Indicators, 2008

The reasoning behind this emphasis is based on two pedagogical facts. The first is that it's impossible to teach all the facts and theories of a typical scientific discipline like astronomy or geology. It's pointless to make students memorize information that they will forget as soon as the class is over, Instead, as the argument goes, we need to teach students to understand how evidence is gathered and how it becomes fact. Teach students how to appreciate science and its power to create knowledge. That's something that will stick with them all their lives.

What are American primaries?

I'm a Canadian who's always been puzzled about American primaries. It seems to me that the purpose of these primaries is to help a political party choose its candidates for the next election. It seems like two of the parties, the Republican party and the Democratic Party, have managed to get state governments to fund their primary elections for reasons that are not very clear to those of us who live in other countries.

Today I was watching Michael Smerconish on CNN. He always has a poll question that provides a deep (and troubling) insight into his way of thinking. Today he announced that he is part of a class action lawsuit demanding that independent voters be allowed to vote in primary elections. That sounds weird to me because I'm used to a system where only members of a party get to choose who their candidates will be.

I was aware of the fact that many Americans see this differently and I knew that some states allow non-party members to pick the party candidate. Nevertheless, I was curious to see how CNN listeners would respond to his poll question.

Here are the results.

I find that result astonishing. 86% of respondents think they should be able to choose the candidate of the Republican or Democrat party even if they don't belong to one of those parties. What do they (you?) think is the purpose of primaries in the United States?

Here's a list of states and who they allow to vote in one of the primaries. It seems like the states actually have laws governing how political parties are able to choose their candidates.

I don't know of any other democracy that has such a bizarre system. I'm a member of one of the political parties in Canada and I participated in selecting our party leader. I would be outraged if my government passed a law allowing members of another party (or nonmembers) to help select my party leader or candidate. Why are such laws acceptable in the United States?

---

Answers in Genesis uses the latest DNA research to destroy evolutionary proof (not!)

There's been so much bad news this week that I though you might enjoy a little humor to lighten your day. Here are some devout Young Earth Creationists making fun of some stupid comments they've found on the internet and calling on some professor to "destroy" evolutionists who believe in junk DNA [Latest in DNA Research Destroys Evolutionary “Proof”].