From jillian.maddox alumni.unimelb.edu.au Tue Apr 2 19:40:46 2013
From: Jill Maddox <jillian.maddoxalumni.unimelb.edu.au>
To: Multiple Recipients of <crimap-usersanimalgenome.org>
Subject: Re: CRI-MAP for multiple, small families?
Date: Tue, 02 Apr 2013 19:40:46 -0500
Hi Layla
The problem will probably be similar for most genetic linkage programs as
it is a property of your data set.
Your number of informative meioses appears to be on the small size for SNP
data especially when you have no phase known data (only 2 generations
genotyped). CRI-MAP works best when there is phase-known data that comes
from genotyped grandparents.
The maximum lod you can get for 60 co-informative meioses (30 pairs
of chromosomes) is 18.06 when theta is 0. Traditionally a lod of 3
has been indicative of linkage (although the correct value used
should vary depending on the number of informative meioses in the
data set). A lod of 3 will only be achieved with fully informative
data if there are at most 15 recombinations (theta = 0.25). If only
one of your families is informative then there will be a maximum of
20 co-informative marker meioses (maximum lod 6.06) and a lod
threshold of 3 will only be exceeded when there at most 2
recombinations between the markers. However this best case scenario
applies to multi-allelic markers where the parents have different
alleles rather than to SNP data.
Double heterozygous matings for all 3 families (theoretical maximum
information if able to detect allele origin) should yield matings where
half the progeny are informative (homozygotes) and half uninformative
(heterozygous - unable to identify which allele comes from which parent
with SNPs and no extra grandparent phase information). So only 30
informative chromosomes for SNP data. The maximum lod for this 9.03 when
theta is 0 and a lod of 3 will only be achieved with at most 5
recombinations at a theta of 0.17. If not all families are informative then
markers will need to be closer together to be linked.
Single heterozygous matings for all 3 families will also yield a maximum of
30 co-informative marker chromosomes.
9 Gb/2000 gives an average inter-marker distance of 9 Mb. Depending
on the species the linkage distance expected for this will vary - ~
1.2 cM/Mb for humans, 0.53 cM/Mb for mice, 0.56 cM/Mb for rats
(https://www.ncbi.nlm.nih.gov/pmc/articles/PMC383296/), 19 cM/Mb for
honey bee (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1626635/). I
don't know what your species will be like recombination
distance/physical distance but if it is like honey bees then you will
have problems given the size of your data set.
Your data is likely to have many markers for which only one family is
informative and these will be harder to link. If you categorise your
markers into groups of number of informative families then you can build up
linkage groups with different lod thresholds - build groups with higher
lods with all families informative, then use these groups to probe lower
informativeness markers (only 2 informative families) with a lower lod and
see what else links in, repeat with single informative families with a
lower lod and then see if you can join your groups. If you use CRI-MAP to
do this then it is semi-automatable with MultiMap and find-all-linkage-
groups or with the Monsanto version of CRI-MAP's autogroup module (given
your families are so small you won't have to worry about the bugs in its
crigen family splitting module).
http://www.parisveltsos.com/research/crimap.html
Regards
Jill
At 06:24 PM 2/04/2013, you wrote:
>
>Hello,
>
>I'm attempting to construct a genetic linkage map from 3 families with 10
>F1 each. I am working with SNP data generated from RAD sequencing of a
>species with a very large genome (9Gb; with presumably a lot of duplication
>and junk). I'm currently working with JoinMap 3.0, but the program isn't
>able to form more than a couple of linkage groups containing ~100 loci,
>when I should have 10 linkage groups with thousands of loci. The other
>~2,000 loci are never paired with any significance. I have tried doing this
>for each family separately and for all three families combined but never
>get more than a couple of linkage groups. Also, when I input files
>containing loci of multiple segregation types, it seems like the program
>always groups loci according to segregation type rather than recombination
>frequency. Unfortunately, I'm not sure if I'm dealing with a computational
>problem, something having to do with the species' genome, or a combination
>of the two. The crosses are of two heterogenously homozygous and
>heterozygous diploid parents (recombination in both). I'm hoping to make
>one map using male-informative loci, another map using female-informative
>loci, then use abxab or abxcd markers to link the two maps together. Do you
>think that CRI-MAP is suited for my purpose? Beldade et al. 2009
>(http://www.plosgenetics.org/...oi%2F10.1371%2Fjournal.pgen.10
>00 366#s3) use a similar strategy of multiple small families and were
>successful with CRI-Map. Any suggestions would be greatly appreciated.
>
>Thank you, Layla
>
>
***************************************************************
Jill Maddox 16 Park Square Port Melbourne, 3207 Australia phone: 03 9646
0428 E-mail: jillian.maddoxalumni.unimelb.edu.au
***************************************************************
|
