Last morning talks
I didn't have time to edit these, so here are my notes directly as I jot them down:
An excellent talk by Prezeworski
Looking for recombination events in order to determine whether males are more
data for 728 meioses with many more markers than
own heuristic approach to call markers
to infer recombination events in the father, from pedigree data
cummulative genetics
resolution at which could resolve crossover eventsm 24K crossover
hotspots id'd in analyses of LD (hapmap)
enrichment of crossover events in hotspots
also estimated the fraction of crossovers
looking at hotspots
no diff in the mean of "use of hotspots" between male and female
looked at LRT null: significnbt variation among individuals using a likelihood ratio test
how heritable is this hotspot usage? (given the known pedigree of the individuals)
so individuals are varying significantly in their use of hotspots, and that variation is heritable.
a similar proportion of crossover events occur in hotspots in males and females genome wide
along the genome, there is extensive variation in cross-pver rates at fine scales, supporting inferences based on analyses of LD.
this could in part explain the diff in hotspots btwn chimpanzee and human.
Used the same data to infer constraints on recombination rate in humans.
it's thought that at least one crossover per chr arm is necessary to ensure proper disjunction.
interesting if look at Lenzi et al. female meiosis is error prone. This analysis is impt for evol bio and human genetics.
How often does proper disjunction occur per gamete among viable individuals.
Modeled the distribution of no of crossover events on each chromosome
find that the proper disjunction can occur in the absence of any crossovers on a chromosome.
how strong would chromatid interference have to be to explain this?
alternatively,
is there a back up mechanism in humans as there is in Drosophila?
There's no known affect on relationship of paternal age and no of recombination events
viable offspring of older mothers have more crossover events; Possibly to overcome insults to the meiotic exchange over the 35-40 years of female's lifespan.
Dan Neafsey talked about widespread selection and frequent recombination in the genome of the Plasmodium falciparum malaria parasite
genetic diversity drives this disease: immune evasion, drug resistance, vaccine failure.
tackling diversity
112K known SNPs
1 snp per 200 bp
trying to capitalize on known snps
SNP avg call rate 91%
avg accuracy 91% and good call rate across samples
Showed that there is a geographic population structure (at the continent and sub-continent level)
ML, HKY+ gamma
LD reflects the local epidemiology
plotted LD for the different continental populations: senegal, thailand, brazil
LD up to 100Kb in brazil vs. much lower in senegal- suggesting a different effective population size
incidence of mult infections is highest senegal (more than one genotypic variety)
while recombination rate doesn't vary, outcrossing does since outcrossing requires multiply-infected hosts (mosquitos often only bites one, so not likely to happen there).
selection: pattern of divergence universal pattern. nonsyn > silent
selection is impacting the
freq in the spectrum isn't biasing nonsynonymous vs silent divergence (mapped avg DAF against proportion of SNPs with signiical Fst (P<0.05))
so what's the role of selection in this pattern?
patterns of div within populations (as opposed to btwn pop)
patterns indicates negative selection
purifying selection comes from derived allele frequencies (mapped for Senegal and Thailand) snp population frequence relative to percent of SNPs)
selective sweeps interrupted by crossover events
3 of 33 strains exhibit ancestral allele
but we also know that positive selection plays a role, too
selective sweeps are detectable
looked at the pfcrt locus, saw disparity as expected.
cq resistant
high diversity regions of the genome a
decoupling : (ie, reduced divergence in high diversity regions)
thinks this is the result of balancing selection: retarding divergence
malaria SNP diversity influenced by pos and neg sel
LD is short, del on local epidemiology
new genotyping tools, 75K SNP genotyping array and Taqman molecular barcode.
Jeffares asked if Neafsey is looking for conservative vs radical aa changes, as well as looking for functional enrichment along selection profile.
Then was a talk about the 17q21.31 microdeletion syndrome (see Genome Research May, 2008)
these events are deNovo, estimated that 1% unexplained mental retardation is probably related to this microdeletion.
All the cases identified so far has been european
The inversion is under pos selection; ind which carry polymorphism show inc fecundity and global recombination
relationship btween microdel and inversion?
tested 17 parents of children
looked for the inversion in non human primates using FSH.
They estimated a human duplicated gene inversion toggling
inverted H2 haplotype most likely represents the ancestral state of huymans
inversions represent a premutation state for large scale de novo microdeletion and disease, maybe because of structural variation of segmental duplications favors NAHR.
(all this in Kidd, Nature 2008)
In all, the meeting was VERY informative. Many posters were also worth looking at. I will have the abstract book available in the computational lab.
Saturday, May 10, 2008
Friday, May 9, 2008
The special session on non-human biology of genomes
They tell me that in past years, the Biology of Genomes meeting didn't have such a focus on human genomics nor sequencing methodologies. well, this morning's session was titled "Genomics of non-human genomes"-- hence reduced to a single session, the first speaker (Barsh) had to defend the activity of doing genetics and genomics in non-human species... and I'm thinking, how crazy is that?? This IS Biology of genomes, afterall...
Anyhow, talks of note below--
Work done by Carlos Bustamante et al:
dog human and cow genomes reveal extensive human reorganization of domesticated genomes.
Prompted by large scale projects to document genome-wide variation in many species, they've taken up the CanMap (genetic diff among domestic dog breeds) and GSK Propgen projects (but he didn't talk about the GSK project).
The CanMap project is to genotype 850 dogs and 200 wolves using affymetrix v2.0 array, and use the data to understand demographic history of dogs.
Since each breed fixed for specific phenotypes in a small number of genes, this is easier to delineate than if selection occured independently. Perhaps those same genes can be implicated for the phenotypic changes we see across 'racial' or 'ethnic' lines.
They use joint association and selective sweep mapping where the unit of analysis is breed. They're looking for linked loci or alleles that are major determinants of phenotypes (for example, dog size, see Sutter et al. 2007, Science 316). They perform regression on body size, again at a breed level analysis, and model avg breed body size (using stepwise regression). They then look at predicted size against the observed, for validation and to determine outliers.
They claim that the distribution of phenotypes is not independent of the distribution of how the breeds are mapped into "genetic space" (in other words, how closely related are they). They look for PCA (but he didn't exactly describe what the high-dimensional feature vector is), assuming that the axes you come up with would relate to the phenotypic differences among the breeds.
bottom line: PCA reconstructs aspects of known breed history.
What's the punchline?
In the same way, they do a spatial prediction of ancestry... in HUMANS, and come up with a remarkable geographically associated demographic history. So we can have some insight into human demographic history migration and admixture, by looking at ourselves at the 'breed' level...
The next talk was by Andersson from Uppsala University about pigmentation mutation in horses ("grey") identified by three phenotypes (these are in fact 'white horses' in whom 'greying with age' appears at a very early age):
1. loss of hair pigmentation
2. melanoma (not because of the UV affect- their skin is actually very dark), but because of intrinsic aspects of the mutation.
3. vitiligo
4. speckling and 'blood marks' (pointing to somatic instability? <- this assumption was questioned by the audience)
They looked for the mutation responsible for this phenotype and found the locus to be a duplication in an INTRON, using hi resolution IBD mapping (a duplication that showed correlation with the phenotype). This "greying with age" is a cis-acting regulatory mutation, but its effect on expression isn't exactly clear yet.
With respect to the black spots appearing on many white horses: says that with a duplication you can recruit silencing. Others questioned whether it could be an epigenetic silencing of the alleles (without implying DNA instability).
Also of note was the grand "unveiling" of the Platypus genome sequence. This week in Nature and accompanying pubs in Genome Research.
There were many posters last night that are of direct interest to Plasmodium genomes (notably the work of Dan Jeffares, Jeff Chang, and one more). The abstracts will be available in the abstract book. There was also a talk that pertained directly to Lucia's work in building phylogenies and hypothesizing when/where lateral gene transfer occured-- although this was on Prokaryote genomes (work by Tal Dagan).
I'm checking out. Better go line up for lunch (the meeting was oversubscribed, and there are even overflow flat panel screens out on the rainy foyer, for those not fitting in the warm and musty auditorium...)
Anyhow, talks of note below--
Work done by Carlos Bustamante et al:
dog human and cow genomes reveal extensive human reorganization of domesticated genomes.
Prompted by large scale projects to document genome-wide variation in many species, they've taken up the CanMap (genetic diff among domestic dog breeds) and GSK Propgen projects (but he didn't talk about the GSK project).
The CanMap project is to genotype 850 dogs and 200 wolves using affymetrix v2.0 array, and use the data to understand demographic history of dogs.
Since each breed fixed for specific phenotypes in a small number of genes, this is easier to delineate than if selection occured independently. Perhaps those same genes can be implicated for the phenotypic changes we see across 'racial' or 'ethnic' lines.
They use joint association and selective sweep mapping where the unit of analysis is breed. They're looking for linked loci or alleles that are major determinants of phenotypes (for example, dog size, see Sutter et al. 2007, Science 316). They perform regression on body size, again at a breed level analysis, and model avg breed body size (using stepwise regression). They then look at predicted size against the observed, for validation and to determine outliers.
They claim that the distribution of phenotypes is not independent of the distribution of how the breeds are mapped into "genetic space" (in other words, how closely related are they). They look for PCA (but he didn't exactly describe what the high-dimensional feature vector is), assuming that the axes you come up with would relate to the phenotypic differences among the breeds.
bottom line: PCA reconstructs aspects of known breed history.
What's the punchline?
In the same way, they do a spatial prediction of ancestry... in HUMANS, and come up with a remarkable geographically associated demographic history. So we can have some insight into human demographic history migration and admixture, by looking at ourselves at the 'breed' level...
The next talk was by Andersson from Uppsala University about pigmentation mutation in horses ("grey") identified by three phenotypes (these are in fact 'white horses' in whom 'greying with age' appears at a very early age):
1. loss of hair pigmentation
2. melanoma (not because of the UV affect- their skin is actually very dark), but because of intrinsic aspects of the mutation.
3. vitiligo
4. speckling and 'blood marks' (pointing to somatic instability? <- this assumption was questioned by the audience)
They looked for the mutation responsible for this phenotype and found the locus to be a duplication in an INTRON, using hi resolution IBD mapping (a duplication that showed correlation with the phenotype). This "greying with age" is a cis-acting regulatory mutation, but its effect on expression isn't exactly clear yet.
With respect to the black spots appearing on many white horses: says that with a duplication you can recruit silencing. Others questioned whether it could be an epigenetic silencing of the alleles (without implying DNA instability).
Also of note was the grand "unveiling" of the Platypus genome sequence. This week in Nature and accompanying pubs in Genome Research.
There were many posters last night that are of direct interest to Plasmodium genomes (notably the work of Dan Jeffares, Jeff Chang, and one more). The abstracts will be available in the abstract book. There was also a talk that pertained directly to Lucia's work in building phylogenies and hypothesizing when/where lateral gene transfer occured-- although this was on Prokaryote genomes (work by Tal Dagan).
I'm checking out. Better go line up for lunch (the meeting was oversubscribed, and there are even overflow flat panel screens out on the rainy foyer, for those not fitting in the warm and musty auditorium...)
Wednesday, May 7, 2008
Greetings from CSHL
Greetings from Cold Spring Harbor.
So far the sessions I've attended have comprised of a variety of focus-applications of large sequencing projects.
These have been:
Functional and Cancer Genomics
Genetics of Complex Traits
and
High Throughput Genomics & Genetics
The latter session was really all about new sequencing technologies and large genome sequencing projects. The most news-worthy talk was Chad Nusbaum's presentation of preliminary findings from their research using the SOLiD(TM) System. There were a number of science publications that covered this this morning (for instance here's the CNBC piece), so I won't go into it here.
One of the interesting ideas promoted by EA Grice from NISC comparative sequencing program was to look for a core microbiome. First, is there a core microbiome? In her talk titled "core diversity profile of human skin microbiome in health and in disease," the center is asking whether we should consider the dominant and rare bacterial populations in normal individuals as a taxonomic
unit when we're predicting disease or drug response (given that our drug digestion or response is sometimes dependent on the response of our microbiome). Their approach was all about the "Phylogenetic architecture" of the microbiome. While they were looking at skin microbiome, they said that the results (and ability to differentiate healthy and diseased tissue) could be generalized. However, they did recognize that skin is special because they have terminal differentiation.
I also presented my poster last night. The set-up is a bit weird because they make you put your poster up and down within a 4 hour period. We got some interest from the usual suspects, but I think there was greater interest in J.D. Watson's wine and cheese birthday party taking place on the lawn. So I got to see J. Watson in a cowboy hat. Personal opinion aside, everyone questioned CSHL's love-hate relationship with Watson...
J.D. Watson's wine and cheese birthday party
It was hard for me to see other posters last night since I was attending my own, but a few interesting things I saw:
Adam Siepel's lab is looking at some interesting ways of finding already characterized regulatory motifs under selection in multiple genomes with a phyloHMM approach (fabsCons derivative method)
Manuel Garber, Michelle Clamp and Xiaohui Xie looking for selection signatures. The idea is to identify sites with unlikely substitution patterns using diversion from stationary distribution that is in background. They tried this over the encode regions.
The evening session went until 11:30pm (!) finishing with R. Durbin's talk on the "1000 genome project." Since the meeting is oversubscribed, many of us are staying several miles away (and I didn't bring my car) so I'm dependent on the shuttle. So I didn't get to leave the labs until 11:45, and it wasn't because I hung out at the bar...
This morning's session is on Computational Genomics.
From JGI: Microbial genomes are a rich source of biological info. Doing comparative prokaryote genomics, they identified antimicrobial peptides, small RNAs and new restriction enzymes that may act as barriers to horizontal gene transfer (see Sorek, Science 2007).
From E birney's lab:
Herrero presented Enredo (which we heard about from Birney on his April 22nd visit)
Also introdued Ortheus
addressing inference of insertion deletion histories and substitution events
Uses a multiple alignment as guiding input
subs are using tamura nei nucleotide subs model
ancestral sequences are represented using weighted sequence graphs, but works progressively.
so check out
enredo, pecan and ortheus.
at their respective websites (which went past on the screen way too fast for me to catch).
Manolis Kellis talking about comparative genomics of mammalian phylogeny and drosophila.
In mammals: want to discover regions of increased selection, detect func elements by increased conservation, and add more genomes in which we can detect smaller elements, and subtle selection
also: discover diff classes of fun elements, patters of change that distinguish diff types of func element.
(I'm not sure what is new here)
They were promoting the "evolutionary signatures" of protein coding genes respecting frame shifting gaps and synonymous substitution signature (again, not new)
built a model for expected substitution counts (diff codons have diff distribution for synonymous subs counts) and score windows for depletion of syn subs-- but this only looks for purifying selection.
Also looking for signatures for microRNA genes (in Drosophila) (see Stark et al, Genome Research 2007)
So far the sessions I've attended have comprised of a variety of focus-applications of large sequencing projects.
These have been:
Functional and Cancer Genomics
Genetics of Complex Traits
and
High Throughput Genomics & Genetics
The latter session was really all about new sequencing technologies and large genome sequencing projects. The most news-worthy talk was Chad Nusbaum's presentation of preliminary findings from their research using the SOLiD(TM) System. There were a number of science publications that covered this this morning (for instance here's the CNBC piece), so I won't go into it here.
One of the interesting ideas promoted by EA Grice from NISC comparative sequencing program was to look for a core microbiome. First, is there a core microbiome? In her talk titled "core diversity profile of human skin microbiome in health and in disease," the center is asking whether we should consider the dominant and rare bacterial populations in normal individuals as a taxonomic
unit when we're predicting disease or drug response (given that our drug digestion or response is sometimes dependent on the response of our microbiome). Their approach was all about the "Phylogenetic architecture" of the microbiome. While they were looking at skin microbiome, they said that the results (and ability to differentiate healthy and diseased tissue) could be generalized. However, they did recognize that skin is special because they have terminal differentiation.
I also presented my poster last night. The set-up is a bit weird because they make you put your poster up and down within a 4 hour period. We got some interest from the usual suspects, but I think there was greater interest in J.D. Watson's wine and cheese birthday party taking place on the lawn. So I got to see J. Watson in a cowboy hat. Personal opinion aside, everyone questioned CSHL's love-hate relationship with Watson...
J.D. Watson's wine and cheese birthday partyIt was hard for me to see other posters last night since I was attending my own, but a few interesting things I saw:
Adam Siepel's lab is looking at some interesting ways of finding already characterized regulatory motifs under selection in multiple genomes with a phyloHMM approach (fabsCons derivative method)
Manuel Garber, Michelle Clamp and Xiaohui Xie looking for selection signatures. The idea is to identify sites with unlikely substitution patterns using diversion from stationary distribution that is in background. They tried this over the encode regions.
The evening session went until 11:30pm (!) finishing with R. Durbin's talk on the "1000 genome project." Since the meeting is oversubscribed, many of us are staying several miles away (and I didn't bring my car) so I'm dependent on the shuttle. So I didn't get to leave the labs until 11:45, and it wasn't because I hung out at the bar...
This morning's session is on Computational Genomics.
From JGI: Microbial genomes are a rich source of biological info. Doing comparative prokaryote genomics, they identified antimicrobial peptides, small RNAs and new restriction enzymes that may act as barriers to horizontal gene transfer (see Sorek, Science 2007).
From E birney's lab:
Herrero presented Enredo (which we heard about from Birney on his April 22nd visit)
Also introdued Ortheus
addressing inference of insertion deletion histories and substitution events
Uses a multiple alignment as guiding input
subs are using tamura nei nucleotide subs model
ancestral sequences are represented using weighted sequence graphs, but works progressively.
so check out
enredo, pecan and ortheus.
at their respective websites (which went past on the screen way too fast for me to catch).
Manolis Kellis talking about comparative genomics of mammalian phylogeny and drosophila.
In mammals: want to discover regions of increased selection, detect func elements by increased conservation, and add more genomes in which we can detect smaller elements, and subtle selection
also: discover diff classes of fun elements, patters of change that distinguish diff types of func element.
(I'm not sure what is new here)
They were promoting the "evolutionary signatures" of protein coding genes respecting frame shifting gaps and synonymous substitution signature (again, not new)
built a model for expected substitution counts (diff codons have diff distribution for synonymous subs counts) and score windows for depletion of syn subs-- but this only looks for purifying selection.
Also looking for signatures for microRNA genes (in Drosophila) (see Stark et al, Genome Research 2007)
Subscribe to:
Posts (Atom)