An interesting thing happened at the first day of GAW 17 presentation. The group discussing collapsing methods announced that the collapsing methods don’t work. Two questions come out quickly. One is from Qunyuan Zhang (working with Michael Province). He basically asked whether such announcement is proper since so many works about collapsing methods have done and published. We also got pretty good results using collapsing schemes, so I asked another question, did your group members test on pathway level? If not, whether it’s because of the problem of collapsing methods or the genetic units.
Now, it turns out people are interesting on the tests using signaling pathways. Actually we did the tests on SNPs and Genes level, we didn’t get good results except two genes either. We didn’t do type I error and power tests because of the tight timeline. So two groups are interested in asking me to join in now. I may join the group with more signaling pathways based approaches to see whether I can get more helpful information.
The good part of the meeting is that people are working on the same dataset and you can see the position of your work among people. The bad part of the meeting with many junior discussants is that they can make some rather basic mistakes.
It seems most found only FLT1 and/or KDR which can be found using Plink. We detected these two genes significantly as well. In general, our work maybe among one of the qualified ones. People are crowed in front of our poster until the late of the next session.
The rest day is much easier since the similar works are distributed among the different topics. The impressive one is that Robert Elston and one group presented their work using the style similar "are you smarter than a 5th grader?" Another group used cartoon movie which is very interesting. Good to meet and communicate with all types of people who are working with you in the same field.
Sunday, October 17, 2010
Wednesday, October 13, 2010
IGES meeting
IGES is finished today. Try to write down something before I may forget.
IGES oral presentations include many interesting topics including rare variants, secondary phenotype, gene pleiotropy, multiple phenotypes, gene-gene interactions, gene-environment interactions, and signaling pathways. I list some about rare variants and secondary/multiple phenotype discussion.
Though it's still early, it looks that people has started to work on mapping rare variants that are associated with diseases. I noticed that there are interesting and new methods about rare variants. Particularly, a work from David Conti and his student considered two directions when collapsing rare variants. A work from Qunyuan Zhang and Michael Province used p-value based sum which considered both the significance and the direction of rare variants. The work from Joon Sang Lee et al. also tried to identify the optimal pooling size for rare variants.
A secondary phenotype from J Wang and S. Shete considered the problem differently from ours. We and the ones including Danyu Lin considered the secondary phenotype conditional on the primary disease status. However, this work tries to correct the bias that may be brought in if consider the secondary phenotype as either a case or a control. There are also a bunch of studies about multiple phenotypes. This intrigues me to wonder whether the problem of multiple phenotypes should include the problem of the secondary phenotype. Or any inherent connection between them.
People used a lot of linear or logistic regression based models by adding more variants including GXG, GXE and others. A simple mind comes up: can we get better models instead of only logistic or linear models?
People also tried to use Bayesian to do some works. According to my observation, most people used Bayesian is only for the convenience of calculation. The strategies of incorporating prior information are not that mature now.
IGES oral presentations include many interesting topics including rare variants, secondary phenotype, gene pleiotropy, multiple phenotypes, gene-gene interactions, gene-environment interactions, and signaling pathways. I list some about rare variants and secondary/multiple phenotype discussion.
Though it's still early, it looks that people has started to work on mapping rare variants that are associated with diseases. I noticed that there are interesting and new methods about rare variants. Particularly, a work from David Conti and his student considered two directions when collapsing rare variants. A work from Qunyuan Zhang and Michael Province used p-value based sum which considered both the significance and the direction of rare variants. The work from Joon Sang Lee et al. also tried to identify the optimal pooling size for rare variants.
A secondary phenotype from J Wang and S. Shete considered the problem differently from ours. We and the ones including Danyu Lin considered the secondary phenotype conditional on the primary disease status. However, this work tries to correct the bias that may be brought in if consider the secondary phenotype as either a case or a control. There are also a bunch of studies about multiple phenotypes. This intrigues me to wonder whether the problem of multiple phenotypes should include the problem of the secondary phenotype. Or any inherent connection between them.
People used a lot of linear or logistic regression based models by adding more variants including GXG, GXE and others. A simple mind comes up: can we get better models instead of only logistic or linear models?
People also tried to use Bayesian to do some works. According to my observation, most people used Bayesian is only for the convenience of calculation. The strategies of incorporating prior information are not that mature now.
Monday, October 11, 2010
IGES first day impression
The theme of first day workshop is “Next Generation Sequencing (NGS) in Genetic Epidemiological Studies”. Besides academic people, the presenting panel also includes people from companies such as Affymetrix, Illumina, and Gold Helix. The talks are interesting especially the ones from Suzanne Leal and Scott Williams. Scott talks about NGS from evolution point of view. Suzanne’s talk is more related to our works.
Suzanne’s talk basically reviewed the papers what we discussed in Journal Club and our GAW 17 submission abstract including
Cohen JC, Kiss RS, Pertsemlidis A, Marcel YL, McPherson R, Hobbs HH. (2004) Multiple rare
alleles contribute to low plasma levels of HDL cholesterol. Science,305,869872.
Madsen BE and Browning SR. (2009) A group-wise association test for rare mutations using
a weighted sum statistic. PLOS Genet, 5(2), e1000384.
Li B and Leal SM. (2008). Methods for detecting associations with rare variants for common dis-
eases: application to analysis of sequence data. Am J Hum Genet., 83(3), 311-321.
Morris AP and Zeggini E. (2010) A evaluation of statistical approaches to rare variant analysis
in genetic association studies. Genet Epidemiol 34(2), 188-193.
Price AL, Kryukov GV, de Bakker PI, Purcell SM, Staples J, Wei LJ, Sunyaev SR.(2009)
Pooled association tests for rare variants in exon-resequencing studies. Am J Hum Genet., 86(6),
832-838.
Another one she mentioned is her new work named KBAC (didn’t write down the whole name). In this work, the basic idea is to upweight the causal genotype. She also tested it upon gene interaction. I asked two questions here for her new work. (1). How to distinct causal genotypes from non-causal ones, especially in real data, and how distinct it from functional genotypes since she also mentioned that they are different. (2). How did they simulate gene interactions? For (1), she responded that, in simulation, they know whether the genotypes are causal or not. But in general in real data, there’s no way to distinct. For (2), she used some data that are known to involved gene interactions. That is to say, they use the known information.
She also pointed out several challenging problems in this field. I list some here: (1). Population structure/admixture, (2) gene X gene, gene X environment, (3) Signaling pathways, (4) Jointly analyzing rare and common variants, and (5) qualitatively and quantitatively.
Suzanne’s talk basically reviewed the papers what we discussed in Journal Club and our GAW 17 submission abstract including
Cohen JC, Kiss RS, Pertsemlidis A, Marcel YL, McPherson R, Hobbs HH. (2004) Multiple rare
alleles contribute to low plasma levels of HDL cholesterol. Science,305,869872.
Madsen BE and Browning SR. (2009) A group-wise association test for rare mutations using
a weighted sum statistic. PLOS Genet, 5(2), e1000384.
Li B and Leal SM. (2008). Methods for detecting associations with rare variants for common dis-
eases: application to analysis of sequence data. Am J Hum Genet., 83(3), 311-321.
Morris AP and Zeggini E. (2010) A evaluation of statistical approaches to rare variant analysis
in genetic association studies. Genet Epidemiol 34(2), 188-193.
Price AL, Kryukov GV, de Bakker PI, Purcell SM, Staples J, Wei LJ, Sunyaev SR.(2009)
Pooled association tests for rare variants in exon-resequencing studies. Am J Hum Genet., 86(6),
832-838.
Another one she mentioned is her new work named KBAC (didn’t write down the whole name). In this work, the basic idea is to upweight the causal genotype. She also tested it upon gene interaction. I asked two questions here for her new work. (1). How to distinct causal genotypes from non-causal ones, especially in real data, and how distinct it from functional genotypes since she also mentioned that they are different. (2). How did they simulate gene interactions? For (1), she responded that, in simulation, they know whether the genotypes are causal or not. But in general in real data, there’s no way to distinct. For (2), she used some data that are known to involved gene interactions. That is to say, they use the known information.
She also pointed out several challenging problems in this field. I list some here: (1). Population structure/admixture, (2) gene X gene, gene X environment, (3) Signaling pathways, (4) Jointly analyzing rare and common variants, and (5) qualitatively and quantitatively.
Tuesday, October 5, 2010
Touching words
Sometimes feel heart can be tired to get some things through. Occasionally, found pieces of words from a small notebook,
"The world is bound to no man
Little drops of water,
little grains of sand
make the mighty
ocean and
the pleasant land."
Whether I can bound my heart between the sky and the land, and have a moment of peace?... (I have uncertainties to ocean, so I'd rather put me between the sky and the land)
On the corner of the small notebook,
"You are the only one that fills my heart."
Different people have the different reasons to fill the different persons to their hearts. However, I'm happy there is a one for many people so that life can look beautiful sometimes.
"The world is bound to no man
Little drops of water,
little grains of sand
make the mighty
ocean and
the pleasant land."
Whether I can bound my heart between the sky and the land, and have a moment of peace?... (I have uncertainties to ocean, so I'd rather put me between the sky and the land)
On the corner of the small notebook,
"You are the only one that fills my heart."
Different people have the different reasons to fill the different persons to their hearts. However, I'm happy there is a one for many people so that life can look beautiful sometimes.
Tuesday, September 21, 2010
A moment of peace between busyworks
Jammed between works. Deadline came, changed, and working on the final version. I'm done the feedback modification boss brought to me. His meeting is still going on. What can I do now? Other works... too less time for too much work. Lunch... God, forgot the wallet at home this morning. Maybe just a moment of break... so precious. Print another small picture for my baby. Always happy to see his smiling and heartless face.
Sunday, September 19, 2010
So just dance
I feel I am pretty productive recently, though still have pressure to get papers published. It's already Sep. I got one rather good work rejected, mainly because they feel the topic is not so right to their journal. I have another finished work under modification. I have another rejected paper ready to polish. I just finished the work for GAW 17. Of course, I have some other collaborating works going.
It looks everything is busy, but I need to push them, make progress and get things done. Hard work, sadness, mixed of hope. Sometimes, I feel I can't keep things alive, and just want to have a good break. But who can't tell that everyone who keeps good wishes is not experiencing the similar feeling.
A moment of numb. Maybe, I can just ignore the details. Have a moment of peace, with my little sweet man, to listen to the piano. So why not just dance, and maybe you may find you can be just ok or just be wonderful without the need of counting the good and the bad.
It looks everything is busy, but I need to push them, make progress and get things done. Hard work, sadness, mixed of hope. Sometimes, I feel I can't keep things alive, and just want to have a good break. But who can't tell that everyone who keeps good wishes is not experiencing the similar feeling.
A moment of numb. Maybe, I can just ignore the details. Have a moment of peace, with my little sweet man, to listen to the piano. So why not just dance, and maybe you may find you can be just ok or just be wonderful without the need of counting the good and the bad.
Saturday, September 11, 2010
Methods of collapsing rare/common variants for mapping the assocition to traits in next generation sequencing
The time of next generation sequencing (NGS) comes so quick, and all of the sudden all the meetings and journals are ready to welcome for NGS time. This reflects to our research as well. I submitted a grant with transferring the concept of joint analysis of family and unrelated data to NGS. Then we quickly merged to 3 NGS projects.
Non-enough of explaining the most mapped common alleles to common diseases, people swarmed to rare variants. However, some rare variants will explain the disease, some may not since many of them are maybe only the results of evolution. Besides, due to its rare occurrence, larger sample size is needed to draw statistical conclusions. Yet many of the rare variants may occur as random mutations and be rare to repeat in population. All of these made the association study more difficult.
We reviewed the recent papers about rare variants. But most used the simulated data including Li and Neal,AJHG, 2008, Madsen and Browning, PLOS genetics, 2009 and Price et al. AJHG, 2009. The basic idea is that people are trying to upweight the rare variants that have true associations. It also seems that a bunch of GAW17 people are working on the same or similar topic. Obviously, it is worth to dig something here. But I have some little concerns whether it's worthy to draw so much attention?
Non-enough of explaining the most mapped common alleles to common diseases, people swarmed to rare variants. However, some rare variants will explain the disease, some may not since many of them are maybe only the results of evolution. Besides, due to its rare occurrence, larger sample size is needed to draw statistical conclusions. Yet many of the rare variants may occur as random mutations and be rare to repeat in population. All of these made the association study more difficult.
We reviewed the recent papers about rare variants. But most used the simulated data including Li and Neal,AJHG, 2008, Madsen and Browning, PLOS genetics, 2009 and Price et al. AJHG, 2009. The basic idea is that people are trying to upweight the rare variants that have true associations. It also seems that a bunch of GAW17 people are working on the same or similar topic. Obviously, it is worth to dig something here. But I have some little concerns whether it's worthy to draw so much attention?
Subscribe to:
Posts (Atom)