Derek Smith on antigenic cartography

Derek Smith, Professor of Infectious Disease Informatics, University of Cambridge, U.K., has developed a method for visualizing antigenic evolution by creating two-dimensional maps in a process called antigenic cartography. These maps are made with data that provide information on the antigenic properties of the pathogen. In the case of influenza virus, the data come from measuring the ability of an antiviral antibody to inhibit hemagglutination – binding of virions to red blood cells. Such maps show how amino acid changes can affect antibody binding to virus particles, which cannot be done by comparing nucleotide sequences of different virus isolates. By charting influenza virus strains in this way, it should be possible to better understand genetic and antigenic evolution.

I discussed antigenic cartography with Dr. Smith during ICAAC Boston 2010, as part of TWiV 99. View the video below, or right click to download the 292 MB .mp4 file.

TWiV 99: ICAAC Boston 2010

Host: Vincent Racaniello

Vincent tours the 50th Interscience Conference on Antimicrobial Agents and Chemotherapy (ICAAC) in Boston, speaking with exhibitors and visitors, including Professors Derek Smith, Michael Schmidt, Frederick Hayden, and Myra McClure.

Many thanks to Chris Condayan and Ray Ortega of the American Society for Microbiology for recording and editing this episode.

Click the arrow above to play, or right-click to download TWiV #99 (45 MB .mp3, 62 minutes)

Subscribe to TWiV (free) in iTunes , at the Zune Marketplace, by the RSS feed, or by email, or listen on your mobile device with Stitcher Radio.

Links for this episode:

Send your virology questions and comments (email or mp3 file) to or leave voicemail at Skype: twivpodcast. You can also post articles that you would like us to discuss at and tag them with twiv.

Influenza A/Mexico/2009 (H1N1) – Questions & Answers

questionHere are answers to questions sent to virology blog about the new strain of influenza H1N1 that continues to spread globally.

Q: Are you aware of efforts to make the virus strain nomenclature more informative? Understanding what segments/genes are related historically and sequence wise to what is confusing  using the ‘serological’ definitions like H1N1. What about the other segments? What about subtyping further the HA and NA genes?

A: See the question below which refers to efforts to revise nomenclature for influenza viruses. Serological definitions are still used because it is not possible to infer the antigenic properties of the HA or NA from the nucleotide sequence (see antigenic cartography). There is as yet no compelling reason to subtype further. Other genes are not included in the serological typing because these do not play a major role in generating protective antibodies.

Q: Is there a good way to get a statistical handle on the relative rates if  H1N1 isolate detection vs. others currently, particularly co-infection? It will be interesting to see what happens with the flu species population and what the interaction of H1N1 in the background of other species will be.

A: Such information is provided by CDC and WHO; most of us don’t have access to it until it is provided, either on websites  or publications such as MMWR or in research journals. There is no question that this is an interesting time to be observing the interaction of three human influenza A subtypes.

Q: Hello, first time poster but I’ve been reading this blog for some time now.  This may have been discussed here before but I have a very simple question.  The new H1N1 strain – does this represent antigenic drift or shift?

A: What a good question, which no one has asked yet. Let’s look at the textbook definitions. Antigenic shift involves major antigenic changes in which a new HA or NA subtype is introduced into the human population (e.g. H1 to H2). Antigenic drift occurs as a result of point mutations that lead to minor, gradual antigenic changes in HA or NA. Mutations in the HA or NA of human influenza viruses occur at a rate of about 1% per year.

Because the new H1N1 virus does not have a new H/N subtype, it’s not a consequence of antigenic shift. But the new HA is quite different from the HA of last season’s H1N1 virus – at least 117 amino acids, comparing A/Brisbane/59/2007 (H1N1) with A/New York/1669/2009 (H1N1).  Clearly the new H1N1 could not have been derived from the previous H1N1 by a few mutations. But it’s still technically drift.

Q: May I ask you if you feel that the CDC is procrastinating on whether to ask vaccine manufacturers to switch over from the seasonal strain to a swine flu vaccine instead? If and when it does, it will take six months for the first batch to be ready, I read. Some makers have said they can produce both but smaller manufacturers say they can only produce one or the other. My question is why can’t they be produced together as one shot? If the current one is trivalent – H1N1, H3N2 and one other, I suppose – why can’t the swine H1N1 also be added into the same shot? Is there a technical (?antigenic) reason this cannot be done?

A: Please listen to my interview with Dr. Peter Palese on Basically, vaccine production is going forward, in case the new H1N1 strain returns to the northern hemisphere in the fall. There is no biological reason why the vaccine could not be combined with the current trivalent preparation. Obviously keeping the new H1N1 vaccine separate provides flexibility in deciding whether or not to deploy it. If you mix it with the trivalent, the decision cannot be reversed.

Q: Reading a recent ProMed-Mail about the H1N1 flu virus, I read this ” … There are 3 main clusters of H1 subtype swine influenza viruses circulating in pigs in the world, designated as h1.3.2, h1.1.3 and 1.2.5 by a newly proposed universal numeral nomenclature system for all influenza viruses. They correspond to classical, avian-like and human-like swine influenza viruses, respectively” … could you please provide – but in a simple way – basics of this numeral system for nomenclature for flu viruses ?

A: I believe that the ProMed-Mail post is referring to an analysis of influenza virus sequences published in PLoS ONE. In that paper they compared the sequences of thousands of viruses and constructed phylogenetic trees – a graphical representation of the relationships among the sequences. The tree for H1 hemagglutinins is reproduced below. The new proposed nomenclature consists of a letter indicating h (HA) or n (NA) followed by the subtype; then two other numbers based on the branches of the tree. From the image you can see that h1.1, h1.2, and h1.3 represent the three main branches of the tree. The third digit is based on further branching.


Q: I have been to coastal Mexico a few times in April and it has always been hot and humid. One would think these conditions would make virus transmission difficult, so it’s surprising that this outbreak occurred during March and April. I did a bit of digging and I think I know the answer. Since I’ve heard this come up a few times, it may worth sharing this in you next posting. Mexico City is cooler and drier than coastal areas especially during the winter. March and April are among the driest months, and the nighttime temperature is probably cool enough year round. Here’s a chart.

A: Your theory is certainly plausible. But remember, there were also outbreaks outside of Mexico City (e.g. Veracruz). And the outbreak in Mexico seems to be over; the borders of the ‘season’ can be fuzzy. There are always exceptions to the seasonality rule. For example, influenza outbreaks may occur during the winter in closed communities, presumably by contact.

Q: My second question is about minor differences within a strain. I’ve been using some of the HA sequences for the new H1N1 on genbank as a vehicle for learning bioruby, and have noted a few differences. At what point do the differences become significant? On Futures in Biotech 40, Dr. Palese mentioned that 3-4 amino acid differences in the Hemagglutinin protein could be to enough cause an immunological difference. As an example, there is a 6 base-pair difference which results in a 4 amino acid difference between the California (FJ966082) and New York (FJ969509) HA sequences. Is that unusual? BTW, for perspective, I also compared the California sequence to the 1918 H1N1 HA sequence (AF117241) and found 311 base-pair differences (out of 1701), and 79 amino acid differences (out of 567).

A: It’s very difficult to predict antigenicity from the nucleotide sequence, so we can’t say if those differences are significant with respect to protective immunity. See the following from antigenic cartography:

There is a close relationship between genetic and antigenic change for human influenza A(H3N2) virus, but genetic distance can be an unreliable predictor of antigenic distance. For example, a change in a single amino acid may cause a disproportionately large change in the binding properties of a virus strain.

Q: The one thing that I do not understand is the mechanism by which the growth stops and the case count “peaks”. Can you please enlighten me as to the mechanism that causes this? In other words, what is going to stop this virus from infecting tens of millions of people by mid-June and a hundred million a few days later which is the path that it is currently on? How does it actually “poop out” short of infecting everyone.

A: Outbreaks stop for several reasons, including seasonality, and we’ve talked about that here quite a bit. I expect the reduced transmission of the virus in the northern hemisphere will soon stop spread. A second is population immunity. Transmission of viral infections require a certain level of susceptible individuals, which varies for each virus. As the number of immune individuals grows, the virus cannot effectively be transmitted, and the outbreak stops.