- #1
- 24,775
- 792
After i say this rule the first person who says the word "race" has to go home
That is not because it's a bad word, it is a very nice word with nice
etymological roots and a rightful place in the human mind. It simply is not what the thread is about.
------------------
This thread is about Genetic Structure Methodology, as illustrated in the recent article in Science (21 May 2004, vol 304 page 1160)
Genetic Structure of the Purebred Domestic Dog
http://www.akcchf.org/news/press/releases/2004/dogbreeds.pdf
algorithms are modernday monsters. when an algorithm is created we do not know where it will lead. usually to nothing bad though, just to unexpected stuff. We are talking about a genetic structure algorithm.
At present it must be helped along by humans but in principle it could
be entirely programmed. give it 85 blood samples from 85 different breeds of dog, and the unaided computer builds a tree.
it isn't that good yet. It is still very rudimentary. the first tree (see figure 2) had only about 10 breeds of dog, plus a catch-all category
--------------------
with a new technology it is always tempting to point out lots of reasons why it cannot succeed or cannot generalize or will never be able to do this or that. But it's a basically poor idea to do that. Humans are ingenious and you just have to wait and see. Please go along with me in doing two contracdictory things:(1) recognize that a new algorithm is a potential monster that could take us where we don't want to go (2)dont be afraid, just try to see how it works.
---------------------
"We therefore ran structure on the entire data set with increasing values of K...to identify the ancestral source populations."
page 1162
the italicized word structure is the name of a computer program.
when most of us were children the names of computer programs were set in all-caps, these people use italics. It is less jarring and makes for a nicer-looking page. In this article the italicized word appears many time.
---------------------
It is always tempting to be a passive consumer of science. Let "them" do the hard stuff and tell me the results. In this case, strangely enough, the method fascinates me more than the results. The topic, Purebred Domestic Dogs, is both solemn and a bit ludicrous. It is fun but I don't have any urge to take it seriously. the method though.
the next quote is from page 1162
"We first used standard neighbor-joining methods to build a majority-rule consensus tree of breeds (Fig. 2), with distances calculated using the chord distance measure (26)...The tree was rooted using wolf samples."
This sentence has a pleasant clanking sound of contented jargon.
One wants immediately to know what is the "chord distance measure" was it perchance invented by Mr. Cavalli-Sforza? Yes it was, seemingly in 1967.
The revered and still-active gentleman was then 45 years old. It could be his most famous contribution to Dog Genetics.
"...The tree was rooted using wolf samples." This sentence is too beautiful for comment. Some poet writing for the New Yorker might steal it.
--------------
there is a very nice passage in the middle of page 1161. I will try to mouse it so I don't have to type it in myself.
dont you get the feeling that we ought to understand the main steps in the algorithm?
(those like myself who don't yet)
That is not because it's a bad word, it is a very nice word with nice
etymological roots and a rightful place in the human mind. It simply is not what the thread is about.
------------------
This thread is about Genetic Structure Methodology, as illustrated in the recent article in Science (21 May 2004, vol 304 page 1160)
Genetic Structure of the Purebred Domestic Dog
http://www.akcchf.org/news/press/releases/2004/dogbreeds.pdf
algorithms are modernday monsters. when an algorithm is created we do not know where it will lead. usually to nothing bad though, just to unexpected stuff. We are talking about a genetic structure algorithm.
At present it must be helped along by humans but in principle it could
be entirely programmed. give it 85 blood samples from 85 different breeds of dog, and the unaided computer builds a tree.
it isn't that good yet. It is still very rudimentary. the first tree (see figure 2) had only about 10 breeds of dog, plus a catch-all category
--------------------
with a new technology it is always tempting to point out lots of reasons why it cannot succeed or cannot generalize or will never be able to do this or that. But it's a basically poor idea to do that. Humans are ingenious and you just have to wait and see. Please go along with me in doing two contracdictory things:(1) recognize that a new algorithm is a potential monster that could take us where we don't want to go (2)dont be afraid, just try to see how it works.
---------------------
"We therefore ran structure on the entire data set with increasing values of K...to identify the ancestral source populations."
page 1162
the italicized word structure is the name of a computer program.
when most of us were children the names of computer programs were set in all-caps, these people use italics. It is less jarring and makes for a nicer-looking page. In this article the italicized word appears many time.
---------------------
It is always tempting to be a passive consumer of science. Let "them" do the hard stuff and tell me the results. In this case, strangely enough, the method fascinates me more than the results. The topic, Purebred Domestic Dogs, is both solemn and a bit ludicrous. It is fun but I don't have any urge to take it seriously. the method though.
the next quote is from page 1162
"We first used standard neighbor-joining methods to build a majority-rule consensus tree of breeds (Fig. 2), with distances calculated using the chord distance measure (26)...The tree was rooted using wolf samples."
This sentence has a pleasant clanking sound of contented jargon.
One wants immediately to know what is the "chord distance measure" was it perchance invented by Mr. Cavalli-Sforza? Yes it was, seemingly in 1967.
The revered and still-active gentleman was then 45 years old. It could be his most famous contribution to Dog Genetics.
"...The tree was rooted using wolf samples." This sentence is too beautiful for comment. Some poet writing for the New Yorker might steal it.
--------------
there is a very nice passage in the middle of page 1161. I will try to mouse it so I don't have to type it in myself.
dont you get the feeling that we ought to understand the main steps in the algorithm?
(those like myself who don't yet)
Last edited by a moderator: