Genome wide association studies (GWAS) Meta-analysis
1
1
Entering edit mode
3.5 years ago

Hi,

I want to do meta-analysis (according to my research goals) by summaries of 60 articles. Due to absence of database for GWAS of livestock, can this meta-analysis be done by the statistical summaries available in the articles? Is the PLINK software capable of summaries analyzing of 60 articles? What software do you recommend for GWAS meta-analysis in estimating models with fixed and random effects?

meta-analysis Plink • 2.2k views
ADD COMMENT
0
Entering edit mode

Hello,

Thank you so much for your reply,

I have access to the parameters of the ID SNP, BTA, GeneName, Position, MAF, Estimate, SE, P-value for GWAS meta-analysis, but I do not have OR in some articles. Is there a way to calculate it? Can something be replaced?

ADD REPLY
0
Entering edit mode

Please use ADD COMMENT/ADD REPLY when responding to existing posts to keep threads logically organized. This comment belongs under @Kevin's answer.

ADD REPLY
0
Entering edit mode

If you have the allele tallies, you can easily calculate odds ratios: A: SNP dataset and Z Score

The PLINK meta-analysis tool requires PLINK-formatted output, so, headers would be along the lines of:

CHR  SNP  BP  A1  C_A  C_U  A2  CHISQ  P  OR  SE  L95  U95

I doubt that you have that for every study, in which case your best option may be METAL.

Alternatively, wait around for the PLINK developer to log in to follow up.

ADD REPLY
0
Entering edit mode

Hello Mr Blighe

I do not have a case and control in my study. I work with quantitative traits. Is there a way to calculate OR for quantitative traits?

ADD REPLY
0
Entering edit mode

So, if not case / control, what is your outcome variable, i.e., your phenotype?

ADD REPLY
0
Entering edit mode

Yes, phenotype traits.

ADD REPLY
0
Entering edit mode

Cool. What are you ultimately aiming to do? - use genotypes to predict the levels of continuous traits / phenotypes?

Can you confirm what is the BTA column in your data? - this is likely the beta coefficient. If you obtain the exponent of this, then you can have the odds ratio. For example (in R):

exp(BTA)

If they are the beta coefficients, then there are likely both positive and negative values in the list.

Can you confirm?

ADD REPLY
0
Entering edit mode

I want to work on the GWAS meta-analysis in QTL mapping for phenotype traits. Using systematic overviews such as meta-analysis, by aggregating existing information, increase the accuracy of the results of genomic predictions and QTL mapping.

To work with METAL, headers mentioned in METAL doucumentation :

MARKERLABEL SNP

ALLELELABELS RefAllele NonRefAllele

PVALUELABEL P-value

EFFECTLABEL Effect or OR

I do not work with the database and collect information manually. Because I did not find any database associated with GWAS livestock for research that has been done so far.

And I only have this information:

SNP

ALLELELABELS RefAllele NonRefAllele

BTA

Position

MAF

SE

P-value

I just do not have OR or Beta. I think the odds ratio in estimating models with fixed and random effects is necessary. Of course, METAL estimates only fixed effects.

If I approve the BTA column in the R software, does it respond? Does METAL not give errors despite these values (positive and negative)?

ADD REPLY
0
Entering edit mode

Can you point me to one of these studies?

I am not saying that the BTA values are the ORs. I am saying that they may be the beta coefficients (based on the name) - it is your job to find out to what this column means. Obtaining the exponent of a beta coefficient produces the OR.

ADD REPLY
0
Entering edit mode

Hello,

Thank you so much for your reply,

For example: (SNP name: rs41646593, BTA:1, Position (bp): 2515412, Gene name:MIS18A)

Here BTA = 1, as you said: Can you confirm what is the BTA column in your data? - this is likely the beta coefficient. If you obtain the exponent of this, then you can have the odds ratio. For example (in R):exp (BTA)

So I got in R this value to BTA: exp (1)= 2.718282

Is this the same beta coefficient or OR ?

ADD REPLY
0
Entering edit mode

What are some of the other BTA values? What is the range of these?

ADD REPLY
0
Entering edit mode

In my first study, I have these BTAs:

exp(1): 2.718282, exp(3):20.08554, exp(5): 148.4132, exp(8):2980.958, exp(9):8103.084, exp(11):59874.14, exp(14):1202604, exp(18):65659969, exp(19): 178482301, exp (21):1318815734, exp(23):9744803446

Ranged from 2.72 to 9744803446.

ADD REPLY
0
Entering edit mode

Can you direct me to this first study so that I can take a look? The description of the results should be in the manuscript, somewhere.

ADD REPLY
0
Entering edit mode

In Case-control study, effect size is mentioned in the terms of OR(odds-ratio) and in quantitative traits, effect size is mentioned in the terms of BETA. So if you have beta column then it is your effect size. If you use plink for GWAS then these are the terms for effect size.

ADD REPLY
0
Entering edit mode
3.5 years ago

It depends on what exactly is available in these manuscripts? - p-values and odds ratios? You will likely have to devote some considerable time in formatting the output statistics from these, which is fine of course.

plink --meta-analysis could handle this for you. Take a look HERE

METAL is another popular program.

Kevin

ADD COMMENT

Login before adding your answer.

Traffic: 1193 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6