Replications of eye_quantitative_traits hits Genome-wide significant :

For genome-wide significant hits in our respective cohorts for which we (reciprocally) are seeking replication in each other cohort, lead for follow up and publication would be given to the group seeking replication. Effect size , P-value for association at the specific SNPs and method of analysis would be requested (ALSPAC's GWA analyses are ongoing). All work and analyses would be acknowledged in the publications and the replicating group would provide authors for the papers written by the leading group, whatever the results.

The summary of the measures used in our analyses in the 2 croatian population-based studies are

attached with this proposal. We are seeking data in the ALSPAC study for one SNP which reached genome wide significance in our metaanalysis of ocular axial length in the 2 Croatian isolates,and showed the same magnitude and direction of effect in our 2 independent isolates. We would also like, if the meta-analysis proposal (below) was not to go ahead , to get similar data on 2 additional regions (2 SNPs) that showed strong evidence of association with axial length, in females only. These do not reach genome wide significance in the initial genotyped set but show strong cluster of association in the imputed dataset and make biological sense.

Meta-analysis on GWA data for eye traits on a collaborative basis:

This analysis required harmonisation of the traits analysed before pooling data. Each group would analyse his own dataset. We can include more trait data than in the replication study, to maximise the value of the analyses for each group.Material tranfert agreements would be drawn to insure that the shared analysis results are not passed on to third party without formal approval. We propose to use rank normalised data post adjusted for age and sex for axial length and refractive error. This is because spherical equivalent usually has a non normal distribution. We propose to also perform the analysis separately for each gender. Analysis would be done using an additive genetic model, for example using the function formetascore of the R package GenABEL for easy merging of the data.

The output files to be exchanged and meta-analysed lay-out would be :

(1) " name": SNP rs number

(2) "chromosome"

(3)"position"

(4)"strand" ideally all standardise to the top strand using build 36

(5) "allele1"

(6)"allele2"

(7)"build"

(8)"effallele" allele for which effect is reported

(9)"effallelefreq"

(10)"n" number of individuals with genotypes and phenotypes available for that SNP

(11) "beta"

(12) "sebeta"

(13)"p" uncorrected P-value for the additive test

(14)"pgc" above corrected after genomic control correction

(15)"lambda" estimated inflation factor (genomic control lambda) for the test

(16)"pexhwe" exact P-value for HWE test

(17) "call" Call rate for the SNP

Results from this meta-analysis would be described in a paper/papers with shared authorship between the groups and 2 co-corresponding authors, one from each group.