All language subtitles for 57031x_PR_LOD_Scores_04_Calculation_Interpretation_-en

PETER REDDIEN: So let's get a LOD score.

OK, so we'll put in our data.

OK, so log base 10 of 0.75 over 2 cubed times

0.25 over 2 for the first power over 0.25 to the fourth power.

OK, so we get a value here of 0.23.

OK, so now to your question, what

do you do with the LOD score, this

is a statistical test we're trying

to utilize to assess linkage.

And by convention, a threshold for a LOD score

has been set such that if your score is above that threshold,

your probability of concluding linkage incorrectly

is less than 0.05, OK.

And that threshold is a LOD score

of greater than or equal to 3.

OK.

And that threshold is set-- it will depend depending

upon the size of the genome, the number of chromosomes and so on

across species.

But this is a threshold that's been

set by convention corresponds to about 1,000 to 1 odds

or exactly 1,000 to 1 odds.

This is a threshold that has been set that will give you

significance for linkage.

OK, so what do we conclude with our LOD score, the 0.23?

Can we conclude that they are linked?

No, so we don't have enough data.

So we cannot conclude linkage.

We tested for linkage at this theta.

We could test at other thetas, and you get a different LOD

score.

But we selected a theta that gave us

our best chance of seeing linkage,

because we used a theta based on the data.

So we're not going to do better if we try other thetas.

Now you can see that this was a pretty large number

of offspring, but we're nowhere near our threshold of 3.

So how would we ever get significant linkage

for our markers in a diseased gene, OK.

What do you guys think?

You could go around testing families

every time you see a disease crop up,

and you're often going to be in this situation.

So how would we ever get a significant statistical test

here with data that would allow us to move forward and look

at a particular region of the chromosome

where we think our gene is?

Any ideas moving beyond the situation of this problem

and just thinking of general?

Do it across multiple families.

OK, so that's what's done.

So now you can add the data from families.

Because of the property of logarithms,

you can just add the LOD scores from individual families,

or you can include all the informative meiosis

in one individual calculation, whichever you prefer.

OK.

All right, so that's what's done.

Now there's one more thing I want

to mention before going to part two, which

is phase unknown, which is I want to emphasize something

I just mentioned once we went through this,

but you want to use the data from any informative meiosis

across the pedigree.

OK, so we use want to use all the generations

in your pedigree that have informative meiosis.

You might see a generation at the end

and, sort of, your attention get focused on that.

But if the pedigree is long, there

could have been informative meiosis

higher up in the pedigree if you have data

from those generations.