Wednesday, April 10, 2013


the new influenza strain "H7N9" that was first seen in Shanghai 2013
is likely a (46+123578)-reassortment of a waterfowl H7N9 virus and a local
poultry H9N2 virus.

The number of differences in the 8 segments to the best public
known matching poultryflu segments is:
(18+33+16+100+19+150+13+7) in the coding region of
(2282+2276+2153+1685+1499+1400+984+840)=13119 nucleotides,
with an estimated
nucleotide-differences from the next best available poultry matches
in the segments.

The number of differences in the 8 segments to the best public
known matching waterflu segments is:
(80+170+150+67+65+50+80+50) in the coding region of
(2282+2276+2153+1685+1499+1400+984+840)=13119 nucleotides,
with an estimated
difference from the next best available mallardlike matches in the segments.

That gives a (rough,+-20%) estimate for the MRCAD of novel H7N9 

with poultry H9N2 in the segments of

This will probaly go up when new related poultry H9N2 become available now. 
The best whole-virus match is
A/brambling/Beijing/16/2012/11/07(H9N2) with
(17,30,18,708,39,583,20,20) nucleotide differences
That brambling presumably got it from poultry.

This suggests a reassortment getting segments 4,6 from wild birds
(mallardlike) and segments 1,2,3,5,7,8 from poultry (H9N2)

In theory the 46-reassortment could have happened in 2006 and all
the other segments were acquired later by subsequent reassortments,
but that looks unlikely - how would it hide so long, why so many
reassortments? More likely that the common ancestor of the brambling
and H9N2 in 2011 was entirely H9N2.

The bigger mimimum distance in segments 4 and 6 could be explained
because more H9N2 poultry viruses than mallardflu H7N9 viruses
are available.

The (amino-) distance of the new H7N9 to the bird index
in the 8 segments is: (8,13,19,-,9,-,27,34).
This is already very non-mallardlike and away from the index.
I won't expect normal mallard-like waterfowl to catch such a virus.


Let's compare it with other strains.
Early H5N1, A/Ck/HK/YU22/2002 ("serotype Z") had (3,8,9,-,5,-,12,12)
differences from the index. It was in ducks (domestic and wild) and geese.
You can see how it "grows" already, especially in segments 7 and 8.
The original H5N1 outbreak in HK 1997 however was mainly a poultry
reassortant in the inner segments : A/HK/156/1997 (20,19,24,-,12,-,13,22) .
It was not prevalent in wild birds and they could wipe it out in poultry.
The reemerging H5N1 in 2002 then was more wildtype-like and
still circulates today.

Some other examples how flu goes from waterfowl into other species
and then starts to acquire amino  acid mutations in the inner segments:

H5N1,Vietnam 2004:(5,6,7,-,5,-,17,16)
H5N1,Cambodia 2012:(18,13,13,-,9,-,20,27)

human H1N1,USA 1918:(8,7,10,-,9,-,7,5)
human H3N2,USA 2007:(33,28,32,-,35,-,28,37)

equine H3N8,1963:(8,12,14,55,15,63,1,10)
equine H3N8,2008:(18,27,40,75,29,75,10,30)


  1. Hello, i noticed a comment you made on the Virology blog some years ago, in which you wrote: "So, flu B jumped from birds to humans 4000 years ago, I read.
    Flu C 8000 years ago and then they kept in humans.
    Flu-A still tries to get a foot in mammals but didn't succeed the last centuries."

    I'm a science journalist and am very interested in finding out more about the evolutionary history of Flu B and Flu C. Could you possibly provide a reference for your comment?

    Many thanks!



  2. that an old paper from 2001, we've got many new sequences
    and insights since then.
    And the calculation doesn't look reliable to me anyway.
    And flu has 8 segments, not just HA.

    I won't repeat that 4000,8000 figure in blogs or forums again.

    aligned with MAFFT, all segments in one pot, I get as smallest
    differences in codon#2 in promille to flu-A in the 8 segments:


    NA1 to other NAs: 416,371,255,306,386,375,314,397

    so segment 1 of flu_B is probably less than 1000 years away
    segments 2,3,5 of flu_B and segment 2 of flu_C are probably less
    than 2000 years away
    for segment 6 of batflu,6,8 of flu_B,1,4,5,7,8 of flu_C
    I can't say much with this method

  3. here they estimate the time of the most recent common ancester
    of all flu-A hemagglutinins at ~1000 years ago

  4. taking that latter paper and concentrating on the 2nd nucleotide in the codons,
    which mutates slower , I estimate
    H7 and H9 , d2=37% , MRCA ~1100
    H9 and H11,12, d2=28% , MRCA~1500
    H1 and H3, d2~33% , MRCA~
    B-HA and H2 , d2 ~42% , --> MRCA ~800
    C-HA and H1 , d2~47% , --> MRCA ~400

    so, flu-B segment 4 separated ~1200 years ago, and flu-C ~1500 years ago.
    I think that's a better estimate than the one from Suzuki,Nei. (4000 years,8000 years)
    However, this assumes a mutation rate as in influenza-A,
    but influenza B is slower (~1/2 of flu-A)
    Taking the average (A-->B) , would give an estimate of ~1800 years ago for flu-B

  5. segment 1 of flu B may have reassorted with segment 1
    of flu-A as late as ~1600

    hmm, maybe I should make a main blog entry from this

    while the H7N9-entry needs to be updated with the new sequences

    almost all H7N9 sequences seem to go back to ~ January
    or February 2013 with some subsequent reassortments, where
    segments from poultry H9N2 were grasped.

    Except S1, the first confirmed case, a 87 year old male who died
    and had an earlier (?), slightly different strain, also
    reassorted with H9N2, but apparently this virus became rare or died. The other one could have spreaded by a superspreader
    and taken over.