Rhiza Labs FluTracker Forum

you can even check this by hand - but it's tedious.

Anyway, here the encoding of nucleotide-triples into amino-acids:


Alanine,Ala,A,4,GCT,GCC,GCA,GCG
Arginine,Arg,R,6,CGT,CGC,CGA,CGG,AGA,AGG
Asparagine,Asn,N,2,AAT,AAC
AsparticAcid,Asp,D,2,GAT,GAC
Cysteine,Cys,C,2,TGT,TGC
GlutamicAcid,Glu,E,2,GAA,GAG
Glutamine,Gln,Q,2,CAA,CAG
Glycine,Gly,G,4,GGT,GGC,GGA,GGG
Histidine,His,H,2,CAT,CAC
Isoleucine,Ile,I,3,ATT,ATC,ATA
Leucine,Leu,L,6,TTA,TTG,CTT,CTC,CTA,CTG
Lysine,Lys,K,2,AAA,AAG
Methionine,Met,M,1,ATG
Phenylalanine,Phe,F,2,TTT,TTC
Proline,Pro,P,4,CCT,CCC,CCA,CCG
Serine,Ser,S,6,TCT,TCC,TCA,TCG,AGT,AGC
Threonine,Thr,T,4,ACT,ACC,ACA,ACG
Tryptophan,Trp,W,1,TGG
Tyrosine,Tyr,Y,2,TAT,TAC
Valine,Val,V,4,GTT,GTC,GTA,GTG
STOP,Sto,},3,TAG,TGA,TAA



hydrophobic:GAVLIMFWP
hydrophilic:STCYNQ,DE,KRH

_________________
no patents on genes, publish the GISAID sequences !

I feel that finding of such a HUGE S/N ratio on birds is an important discovery by itself!
Biological explanation of such a behavior is also very relevant. More than one model could eventually emerge to fit your already large amount of data.

However, I think the consolidation of this interesting and singular feature deserves priority over bio models. I’m wandering which kind of graphic or picture could better represent those findings on a compact image. I guess which kind of statistics would be appropriated to characterize quantitatively this nonstandard phenomenon.

I don’t know if you already tried, but it could be interesting to plot:
1. nucleotide change density versus distance (*)
2. amino acid change density versus distance

(*) An alternative representation would be [changes] density versus time.
This would be a time-dependent representation, while the first would time-independent representation.
If molecular clock for the gene is approximately stable, then both representations show a similar pictures.

This would reduce the 2100 dimensions (degrees of freedom) into a readable 2D picture.
The concept could be illustrated by a comparison of an avian and a mammalian example.
The bird amino acid will show a huge density close to r=0, your “bigger density” in the center.

Thank you for convincing examples! 350 nucleotide differences but only 4 amino-acid differences. WOW!

Query: I wonder if it such a high S/N without substantial number of 4-fold degeneracy sites that
"can have any of the four bases, with no resulting amino acid replacement.”(1) Reid et al. 2002 op . cit. this thread, page 3.

I have a picture !
Only segment 1 so far. In segment 1 we have a clear
separation of American and Eurasian bird- viruses.
In other segments it's not so clear.

http://magictour.free.fr/panflu/sepp1c0.JPG



no-time graph for segment 1


the biggest nucleotide differences are between pairs, where one PB2 is
American and the other Eurasian.
But for amino acids we have "overlaps" on the continents.

The amino-acid variability seems to be limited (in birds, not in humans)
and the similar amino-acid sequences
come back after some evolution, even on other continents
and separate evolution.




fewer mutation in segment 2, the segment with PB1-F2 and which
jumped from birds to humans in the 1957 and 1968 pandemics

http://magictour.free.fr/panflu/sepp12.JPG

_________________
no patents on genes, publish the GISAID sequences !

Pictures are quite expressive and clear.

Could you add more detailed captions?
Picture a. (top ) is clearly avian, segment 1. Is the “index” sequence at the origin (0,0)?
Green line looks like a scale. Is it, or a feature pointer?
Pictures b,c,d are all avian, OK (which segment?). Any mammalian flu evolution among them?

Although I can clearly see distinct density clusters on picture a, I cannot clearly identify separation of American and Eurasian bird- viruses as clearly as you already do! Please, detail a bit your description. At any rate, I appreciate the graphic quality and the representation effort!

PS What is the scale factor between nucletide and amino acid? I cannot be 1.0, since thare are fewer (much fewer ) amino acid changes than nucleotide changes. The "slope" of these "density clouds" is related to S/N ratio. (S+N)/N, to be precise. If S>>N then (S+N)/N=~S/N.

I was not sure if my “slope” description was clear enough.
I figure out it would be clearer to draw lines roughly on the center of each “density clouds” I could identify by inspection.
Marked slopes lines in blue. No math fit, rather visual guideline. Is the blank gap between two upper “clouds” the separation of American and Eurasian bird- viruses?

Not sure I am correctly reading your picture:
The remark “pairs of avian influenza viruses” suggests that each point is corresponds to an alignment and comparison of two avian sequences, such as you previous example ( A/duck/Altai/1285/1991/08/15(H5N3) segment 1 versus A/pintail/Alaska/779/2005/08/20(H3N8).

The correlation is performed over a wide universe of PAIRS of avian flu isolates. In that case, the picture does not directly represent flu evolution from a progenitor in the center and the evolving children on the circumference, expanding with time. It is mainly a time-independent representation.
Although pictures clearly shows elongated clouds, the interpretation of the marked “slopes” could not necessarily reflect the S/N ratio along evolution.

Regarding the "H5N1 since 1997" picture, is it possible to discriminate the emergence of the 3 BP deletion strains such as recent H5N1 cases in Egypt?

yes, I should add some scale etc. but it's tedious, bif file, slow program.
And you can see the trend
already. The amino-acid differences go to 0 at the y-axis, where the
nucleotide-differences are still high !


no mammalean viruses yet in the pictures.
bottom left pixel is 0,0
1st pic is all segment 1

scale is
x:amino*9: 0-640
y:nucleo*1.7:0-350
the window has 640*350 pixel


the "clouds" are pairs of viruses, the top cloud is one American PB2 and one Eurasian PB2
Some few viruses were deleted , recombinants or distant South-America

groupings are here http://www.flugenome.org/show_subtypes.php

figure out how the groups corresponds to the clouds ... I can't yet easily do it,
the program is slow

_________________
no patents on genes, publish the GISAID sequences !

from Egypt we usually only have HA and NA , segments 4 and 6.
It's similar to Qinghai-H5N1, the picture is for decades,
you won't see yearly differences

_________________
no patents on genes, publish the GISAID sequences !

you are faster with the questions than I can handle it...

> Marked slopes lines in blue. No math fit, rather visual guideline.
> Is the blank gap between two upper “clouds” the separation of
> American and Eurasian bird- viruses?

it separates (Am,Eu) - pairs. Where the (Am,Am) and (Eu,Eu) pairs
are - I don't know yet. They may well form one common cloud
or several clouds for substrains

> Not sure I am correctly reading your picture:
> The remark “pairs of avian influenza viruses” suggests that each
> point is corresponds to an alignment and comparison of two avian
> sequences, such as you previous example
> ( A/duck/Altai/1285/1991/08/15(H5N3) segment 1 versus
> A/pintail/Alaska/779/2005/08/20(H3N8).

yes

> The correlation is performed over a wide universe of PAIRS of avian flu
> isolates.

~3000 viruses, only 1% of pairs are chosen randomly.
I should reduce the list and eliminate similar viruses

> In that case, the picture does not directly represent flu evolution
> from a progenitor in the center and the evolving children on the
> circumference, expanding with time.

right. Time is being ignored. But nucleotide differences usually accumulate
well over time. Better in humans - in birds there are sometimes
strange periods of slow evolution even on the nucleotide-level. But let's
ignore that for now.

> It is mainly a time-independent representation.
> Although pictures clearly shows elongated clouds, the interpretation
> of the marked “slopes” could not necessarily reflect the S/N ratio
> along evolution.

aehh, what's a slope ?

_________________
no patents on genes, publish the GISAID sequences !

average number of amino-acid differences from the index
in avian flu viruses over 5-year periods ,
PB2,PB1,PA,NP,M1 and average

1975 3.62 4.94 4.28 3.31 2.37 370
1980 6.54 3.87 7.53 6.05 3.10 542
1985 5.46 3.29 5.97 5.51 3.32 471
1990 7.59 5.29 8.66 5.56 3.25 607
1995 10.34 9.01 11.11 7.47 3.77 834
2000 8.90 8.53 11.05 6.25 5.50 805
2005 7.70 7.00 9.70 6.70 6.51 752

does it increase ? Thre was much H5N1 in 1997-2006
which is rather distant. And H9N2 recently.
The Canadian mallards sampled in the 70s were close,
not much from Eurasia in that period.

----edit------------

average number of amino-acid differences in segment 1
between North American and Eurasian viruses
in 5 year periods:

1975: 7.87
1980:12.67
1985:10.88
1990:13.49
1995:17.79
2000:14.29
2005:14.34

_________________
no patents on genes, publish the GISAID sequences !