Saturday, February 1, 2014

What’s wrong with population genetics? An example how things can go wrong



This question is provocative, but still rational and defends it's existence.   If something looks like being wrong despite of being science, your bells should ring. It is your responsibility.  The  world trusts and believes that science must be based on objectivity and evidences, not on subjective visions.  If something is in contradiction with other observations which are earlier proved scientifically, it is a reason to be suspicious until the contradiction is resolved.    The population genetics should follow these rules.  Unfortunately today this is not true, the population genetics is sometimes a confused bunch of  subjective visions and doesn’t follow healthy logic where today is before tomorrow, yesterday was before today.  

Expressing this kind of claims needs evidences.  I can prove my claims by methodological way and by facts.   Because methods are often a bit complex and need more learning and special terminology I‘ll present now only simple facts based on data and point out where things go wrong.  Anyone knowing something behind these facts can see problems and everyone is free to show my possible mistakes.   I would appreciate it.  Unfortunately I believe that the discussion will be negligible, because the population genetics is in certain cases already so deep in it’s own swamp that it can’t help itself.    

Here are two principles to follow:

1  Everyone knows that if we derive something from something we must have a correlation between these somethings.   Everyone, even a child intuitively knows that there is a basic law of causation between events if they are in some relation.   The causation obliges that things are bound to time; something happens before and leads to sequential events.  

2  Everyone knows that statistics are based on sampling.   Everyone also knows that the sampling in statistics must be comprehensive and representative.   It cannot be biased.  If we want to make statistics of plants in certain forest, we should sample the whole forest by some method, not only a small part of it, like a small part near a meadow.  

You think that of course scientists know this and do their best.  Perhaps, maybe not.  Maybe they do their best, but then they don’t know what they are doing.   They break both principles I listed above, not a little but they break these rules brutally.   After doing this they can’t go back, thanks for the failed scientific peer reviewing, there is no way to go back.  

So what is so badly wrong that I have reason to worry about the prowess of scientists?  Keeping in mind two principles mentioned above I can present following peculiarities.  Remember that my case is only my observations, somebody else can find other peculiarities.

1  A half of Finnish samples represents a settlement that is around 300 years old and represents a new genetic structure.  Finnish old settlements, representing around  70% of the whole Finnish population, have over 1000 years old roots (defined by written history), over 1500 years old roots (defined by archaeology)  and defined by uniparental genes (I1 and N1c1) thousands years old roots.  Scientists use a younger population to define older populations.  This is the causation error I found in many studies.  

2   Around 50% of used Finnish samples represent only 0,32% of the Finnish population, living far away from the population centers.   It is like using a small village from Sicily to represent 50% of  Italians, including northern Italians.  This means that samples from a sparse populated area is overrepresented by a factor 150!   This is the sampling error I found in studies. 

Maybe someone finds it possible to defend the causation and sampling errors by saying that it is more interesting to see the population diversity, but this is only a desperate way to defend an unscientific personal opinion, although a predictable way.   Following the logic like this we could as well gather samples from Helsinki suburbs where 20-30% of residents are immigrants with Finnish citizenship.   Looking from the statistic perspective the error is same, only the place and the population age differs.  The only acceptable way is to use statistic rules and include minorities if they pass these  rules.    

  

Conclusions 


This what I found is unbelievable.  Although I am not going to handle methods, like admix and PCA analyses, it is clear to everyone who has digested these methods that this error in sampling will affect studies flattening, impoverishing and distorting also results obtained by other samples.  

I can’t understand three things:  

-        -   how Finnish researchers have released these samples taking into account the use for representative purpose of Finns in worldwide studies

-         -  how researchers making these studies had not even single doubt about this problem which could destroy their studies

-         -  how the peer reviewing didn’t reveal anything.  Obviously no bells rang.  Why?


Background


Finnish settlement sampling in studies (late settlement or drifted population in studies)

Until the 17th century, the area of Kuusamo was inhabited by the semi-nomadic Sami.

  • From the 15th century Finnish fishermen also took advantage of fishing grounds on the lower reaches of the river Iijoki near Kuusamo. They took regular trips of a few weeks from Kuusamo, but because the land could not provide hay for cattle elsewhere than near the river, they founded no fixed settlements. Only when, in 1673, the Swedish government granted to all settlers in Lapland a tax exemption for 15 years, settlers from Savo and Kainuu did settle in Kuusamo.


  •  The first parish in Kuusamo was founded in 1685. In 1687 a temporary chapel was built, in 1695 the first church. From the end of the 17th century the area around the lake Kuusamojärvi began to be called Kuusamo. The precise etymology of the name is unclear, however, one possible derivation is from a Sami word for "spruce forest".

Kuusamo population 2008: 16 779




2  Finnish population, age and location

                     Finland population 2014: 5 450 614

Population density today:

 
Today around 70% of the Finnish populations lives in old settlements, most of the rest in the cities of late settlements.  Only a few percents live outside these areas.  Old regions are  Ahvenanmaa, Etelä-Karjala, Etelä-Savo, Kanta-Häme, Kymenlaakso, Pirkanmaa,  Etelä-Pohjanmaa,  Päijät-Häme,  Satakunta, Uusimaa, Varsinais-Suomi.


                      Settlements during the Iron Age:



Agricultural regions around 1000AD (brown and red spots):




















Finnish settlements 1540AD according to old tax catalogues (each small black spot represents 50 houses, every house around 5-10 occupants):











 

Examples of studies using biased Finnish data:


Olalde et. al 
Derived immune and ancestral pigmentation alleles in a 7,000-year-old Mesolithic European


Khrunin et al.
A Genome-Wide Analysis of Populations from European Russia Reveals a New Pole of Genetic Diversity in Northern Europe



Nelis et al. 
Genetic structure in Europeans 




No comments:

Post a Comment

English preferred, because readers are international.

No more Anonymous posts.