HOME

Sample Discussion

Subject: Sheep small SNPchip development

From salamon.d_at_gmail.com  Tue Jun  8 09:14:00 2010
From: "Dragica Salamon" 
Postmaster: submitted from web posting form
To: Multiple Recipients of AnGenMap
Subject: sheep small SNPchip development
Date: Tue, 08 Jun 2010 09:14:00 -0500

Hello, everyone!

As I am new in this discusion/information group, I was wondering if this
would be a good place to talk about the development of SNP chips...
(If not, could you, please, direct me to some discussion groups) 

The thing is, I am in progress of preparing a study on (10-15 indigenous)
eastern Adriatic sheep breeds (diversity, distances, population stuff...)
and I am interested in adopting the new SNP marker methods instead of
microsatellite ones in case I can make my financial frame to fit.

Do you think using 100 SNPs in order to exchange for 30 microsatellites
and using the additional 30SNPs more (of some production and functional
traits) would make an interesting chip construction to share among
researh teams?

I would be happy if to hear back from you...

Dragica, Croatia

From Paul.Boettcher_at_fao.org  Tue Jun  8 09:29:13 2010
From: "Boettcher, Paul (AGAG)" 
To: Multiple Recipients of AnGenMap
Subject: RE: sheep small SNPchip development
Date: Tue, 08 Jun 2010 09:29:13 -0500

Dear Dr. Salamon 

The FAO is also interested to migrate from microsatellites to SNP for 
population characterization. However, we foresee the need for some 
prior research in SNP selection to obtain a representative measure of
diversity.

Paul Boettcher

From taylorjerr_at_missouri.edu  Tue Jun  8 13:20:06 2010
From: "Taylor, Jerry F." 
To: Multiple Recipients of AnGenMap
Subject: RE: sheep small SNPchip development
Date: Tue, 08 Jun 2010 13:20:06 -0500

The Illumina SNP50 assays has been used for phylogenetic work and I suspect
that the existing sheep and swine 50K assays are being used for the same
purpose. See for example:

Decker JE, JC Pires, GC Conant, SD McKay, MP Heaton, K Chen, A Cooper, J
Vilkki, CM Seabury, AR Caetano, GS Johnson, RA Brenneman, O Hanotte, LS
Eggert, P Wiener, JJ Kim, KS Kim, TS Sonstegard, CP Van Tassell, HL
Neibergs, JC McEwan, R Brauning, LL Coutinho, ME Babar, GA Wilson, MC
McClure, MM Rolf, JW Kim, RD Schnabel, JF Taylor. 2009. Resolving the
evolution of extant and extinct ruminants with high-throughput
phylogenomics. Proc. Natl. Acad. Sci. U S A. 106:18644-9.

MacEachern S, McEwan J, Goddard M. 2009. Phylogenetic reconstruction and
the identification of ancient polymorphism in the Bovini tribe (Bovidae,
Bovinae). BMC Genomics. 2009 Apr 24;10:177.

100 SNPs is probably too few for what you want to do. If you can't afford
to use the 50K sheep assay you could develop a 3,072 Goldengate assay that
would cost about $30 per sample if you could afford to genotype 1000
samples. There are likely others out there using the sheep chip for this
purpose and so a great advantage of using this assay is that the data sets
become integrable and extendable.

SNP ascertainment is going to be a problem for estimating genetic distances
probably no matter what assay you use. See the Decker paper about for a
discussion on this.

Good luck,

Jerry

From salamon.d_at_gmail.com  Tue Jun  8 16:00:31 2010
From: =?UTF-8?Q?Dragica_=C5=A0alamon?= 
To: Multiple Recipients of AnGenMap
Subject: Re: sheep small SNPchip development
Date: Tue, 08 Jun 2010 16:00:31 -0500

thank you all on very useful inputs and guidelines in your replys today!

now I have a thing or two to think about for a few days :) like the size of
a chip... the ascertainment bias...

and also to "tame" my somewhat idealistic ideas into, hopefully, start of a
nice and useful project... :)

From jillm_at_rubens.its.unimelb.edu.au  Tue Jun  8 19:02:53 2010
From: Jill Maddox 
To: Multiple Recipients of AnGenMap
Subject: Re: sheep small SNPchip development
Date: Tue, 08 Jun 2010 19:02:53 -0500

Hi Dragica and others interested in sheep SNPs

The International Sheep Genomics Consortium (http://www.sheephapmap.org/)
is currently developing an Illumina Infinium 5K chip with SNPs being
selected from the 50K chip based on their informativeness for multiple
breeds. This chip may be suitable for your needs and given that large
numbers of it will be produced should be relatively affordable. The contact
person for the 5K SNP chip this is John McEwan
. Sheep parentage SNP tests are also being
developed by a couple of organisations (I know of AgResearch/Ovita New
Zealand and Meat and Livestock Australia but there may be others) based on
this information. As far as I am aware the MLA SNP parentage test will be
public domain in terms of what SNPs are being used. I think the AgResearch
SNP list will be private - but John would know more about this.

Regards

Jill Maddox
 
From hillel_at_agri.huji.ac.il  Wed Jun  9 06:47:25 2010
From: "Jossi Hillel" 
To: Multiple Recipients of AnGenMap
Subject: RE: sheep small SNPchip development
Date: Wed, 09 Jun 2010 06:47:25 -0500

Based on our analysis of ten chicken populations (10 birds each) using
STRUCTURE software, microsatellites discriminate between populations far
better than any other marker type. I’d stay with microsatellites for this
kind of need, given the budget is limited.

Jossi Hillel 



From Martien.Groenen_at_wur.nl  Wed Jun  9 07:52:14 2010
From: "Groenen, Martien" 
To: Multiple Recipients of AnGenMap
Subject: RE: sheep small SNPchip development
Date: Wed, 09 Jun 2010 07:52:14 -0500

In the past we have used microsatellites and in recent years moved
completely to SNPs. I would strongly advice to use SNPs rather than micros.
SNPs are more cost effective, allele calling is the same between different
labs and although you need 3-4 times the number of markers, overall the
costs are lower than using micros.

Martien Groenen

From taylorjerr_at_missouri.edu  Wed Jun  9 07:56:08 2010
From: "Taylor, Jerry F." 
To: Multiple Recipients of AnGenMap
Subject: RE: sheep small SNPchip development
Date: Wed, 09 Jun 2010 07:56:08 -0500

This depends on the objectives of the study. If you want to discriminate
between recently diverged populations then running a few microsatellites
and observing 5-6 alleles per locus will do a decent job. There is still
likely to be an issue of ascertainment depending on how the microsats were
discovered and chosen to be run in the populations.

If you are dealing with populations that have much older divergence times
then homoplasy becomes an issue due to the much higher mutation rates of
microsatellites. That is, when you see a specific allele within two
populations can you conclude that they are identical by descent or has
recurrent mutation produced the alleles from parent alleles that were
evolutionarily distinct?

Finally, the logistics of scoring 30-40 microsats is a bit of a pain. Even
if you have these nicely multiplexed you need to process your DNAs 5-6
times which makes them costly and allows opportunity for error (sample
swapping). Simply running one sample one time and producing 50,000
genotypes makes the genotyping part of the project very simple.

The issue of informativeness is simply the sum of the effective number of
alleles at each locus summed across all loci. If you run 30 SNPs the
informativeness is going to be less than if you run 30 microsats. But if
you run 3,000 or 5,000 or 50,000 SNPs you will produce topologies that
effectively have 100% bootstrap support.

Jerry

From hsimian_at_gwdg.de  Wed Jun  9 08:01:44 2010
From: "Simianer, Henner" 
To: Multiple Recipients of AnGenMap
Subject: RE: sheep small SNPchip development
Date: Wed, 09 Jun 2010 08:01:44 -0500

We did a study in eight chicken breeds (8 animals each) to compare
microsatellites and SNPs using STRUCTURE software and principal component
analysis (PCA). For the same number of markers you cannot see any
difference in the PCA-Plot of the first two components or for the
differentiation in the distruct-plot. We also calculated the euclidean
distance and from that a statistic that describes the level of
differentiation. These values show that the differentiation surprisingly is
much better for the SNPs compared to microsatellites even with the same
number of markers. These results will be presented at the WCGALP in Leipzig
in August 2010.

Christian Gaerke and Henner Simianer

From hillel@agri.huji.ac.il  Wed Jun  9 10:58:35 2010
From: "Jossi Hillel" 
To: Multiple Recipients of AnGenMap
Subject: RE: sheep small SNPchip development
Date: Wed, 09 Jun 2010 10:58:35 -0500

Hi again,

In my previous message I pasted a figure into the mail and it turned out
that many of the recepients did not get it. Here I attach that figure from
which one can see informative results. From this figure you can see that 29
microsatellites did much better job than 152 SNPs. If these results are not
in full agreement with those mentioned in previous messages by Martien and
Henner, it is possible that apart from the microsatellites and intergenic
loci in the attached figure, all others SNP markers are within genes or
related to genes. I wonder where the SNPs mentioned by Henner and Martien
are located. These results will be submitted for publication very soon.

Jossi Hillel

From john.mcewan@agresearch.co.nz  Wed Jun  9 14:31:13 2010
From: "McEwan, John" 
To: Multiple Recipients of 
Subject: RE: sheep small SNPchip development
Date: Wed, 09 Jun 2010 14:31:13 -0500

Dear all

The "rule of thumb" for mammalian species is 1 microsatellite has the
discrimination power of about 5 SNPs for a variety of typical tasks

29*5 ~ 152

However, SNPs and microsatellites have different utility depending on the
task. Microsatellites have a higher mutation rate and even with the problem
of recurrent mutations this means they are somewhat better on a blow by
blow comparison of recent fine scale differences on the Y chromosome for
instance.

However, please be realistic here. The cost of a SNP on a 50K chip at
~US$100 for consumables is the equivalent cost of genotyping say 30-50
microsatellites (with multiplexing) but the power of the results is way way
better for almost all tasks and the information that can be extracted just
cannot be compared sensibly. In addition the technology costs are much more
scalable.

John McEwan

From lipkin@ginegar.net  Wed Jun  9 22:30:06 2010
From: "Udi Lipkin" 
To: Multiple Recipients of 
Subject: RE: sheep small SNPchip development
Date: Wed, 09 Jun 2010 22:30:06 -0500

The problem is that you can't have each "low cost" SNP alone. You must take
the entire array, and the cost of this array is sharply limiting your
population size. Thus, with SNPs you have far many more markers per
individuals, but far fewer individuals per marker. As a result, in many
studies the population dimension is lost.

Ehud Lipkin

From Peter.Hunt@csiro.au  Wed Jun  9 22:38:54 2010
From: 
To: Multiple Recipients of 
Subject: RE: sheep small SNPchip development
Date: Wed, 09 Jun 2010 22:38:54 -0500

Dear all,

Have been following this email trail with interest.

Homoplasy was raised earlier in the discussion regarding differences
between SNP and Msats. I have noticed that there seems to be little
discrimination between Ts and Tv SNPs in chip design (at least for the
sheep one). Has anyone considered the higher potential for homoplasic
mutations in Ts compare to Tv SNP? For some investigations a Tv only SNP
chip might be preferable? Has anyone studied Tv data separate to Ts data to
see if there is any divergence of results? Are there enough Tv on the
current chip to make a fair comparison anyway?

I'd be interested in any comments or information on these points,

Peter

--------------------------------------------------------------- 
Dr Peter Hunt 
Research Scientist, 
CSIRO McMaster laboratory, 
Locked Mail Bag 1, 
Armidale, NSW, 2350 
Australia 

Phone +61 2 6776 1300 Fax +61 2 6776 1333 email peter.hunt@csiro.au
--------------------------------------------------------------- 

From taylorjerr@missouri.edu  Wed Jun  9 23:54:43 2010
From: "Taylor, Jerry F." 
To: Multiple Recipients of 
Subject: Re: sheep small SNPchip development
Date: Wed, 09 Jun 2010 23:54:43 -0500

Yes, we have. We found no difference in the cattle tree in the decker et al
paper, but there is a big difference in the numbers of Ts vs Tv SNPs on the
BovineSNP50.

Jerry

Sent from my iPhone

From salamon.d@gmail.com  Thu Jun 10 09:49:05 2010
From: =?UTF-8?Q?Dragica_=C5=A0alamon?= 
To: Multiple Recipients of 
Subject: Re: sheep small SNPchip development
Date: Thu, 10 Jun 2010 09:49:05 -0500

Looks like the SNP discussion has a few new questions and directions :)

as far as my first question from a few days ago is concerned, I found all
of your replies to be very helpful! (both on topic and in priv. mailbox)
Thank you.

I'm more or less decided how to address my research ideas now, ... however
I still have questions on this diversity use issue with the ascertainment
bias...

How do you think this problem could be addressed in analysis of the 5K chip
data, or the Goldengate assay data... Would it be better maybe, or could
this problem somehow be be taken care of by different selection of the SNPs
for the chip construction? Or is the size of the chip, also in this case,
the alleviating factor?

Regards, 
Dragica

From taylorjerr@missouri.edu  Thu Jun 10 10:05:44 2010
From: "Taylor, Jerry F." 
To: Multiple Recipients of 
Subject: RE: sheep small SNPchip development
Date: Thu, 10 Jun 2010 10:05:44 -0500

Dragica:

To overcome the problem, the SNPs need to be considered to be a random
sample of the SNPs (in terms of their minor allele frequencies) from each
of the assayed populations. This is clearly a technically difficult problem
since it would require independent SNP discovery projects which were not
biased toward finding the common SNPs within each population (highly non
trivial and would require a lot of deep sequencing). It also runs highly
counter to the design philosophy of most genotyping assays in which we
deliberately bias the content towards SNPs with high MAF in all
populations.

I have discussed this with several heavy-hitting mathematical geneticists
and am trying to put some collaborations together to see if a mathematical
"fix" might be possible. But for now, all the things I tried (see Decker
paper) were fruitless.

Jerry

From gianola@ansci.wisc.edu  Thu Jun 10 10:21:14 2010
From: Daniel Gianola 
To: Multiple Recipients of 
Subject: RE: sheep small SNPchip development
Date: Thu, 10 Jun 2010 10:21:14 -0500

And probably the sheep are not a random sample either, so there are two
sources of non-ignorable missingness. A better name for would be nonrandom
sheep small nonrandomSNP chip. Please make it cheap too.

Dan

Go back to the AnGenMap main page.

 

 

© 2003-2024: USA · USDA · NRPSP8 · Program to Accelerate Animal Genomics Applications. Contact: Bioinformatics Team