<div>Dear Sodbo, Diego,</div><div><br></div><div>It looks like either 'reshuffle' or CLACK-GWAS do something wrong - and I actually think this is reshuffle (see my tests below). Basically, I run analysis on 'big' genotype set (gtDataChr1) and a thinned one, where I take every 100th marker only (gtDataChr1small). When I extract results with reshuffle, I see a marker with chi2>25 in small, but not in the original big data set!</div>
<div><br></div><div>You can find the test-files in my public dropbox folder (I am going to send the link to you separately; if anyone else is interested in testing, let me know); the commands I used for testing are provided below.</div>
<div><br></div><div>best, and let mw know,</div><div>YA</div><div><br></div><div>Running analysis:</div><div><br></div><div>CLAK-GWAS -cov covData -phi gKinId -pheno phDataChr1 -snp gtDataChr1small -out BAllSmall -var eigen -nths 4</div>
<div>reshuffle BAllSmall --chi=25</div><div>[identifies one SNP-trait pair, rs3738547-GI_7657133.S, with chi2>25]</div><div>CLAK-GWAS -cov covData -phi gKinId -pheno phDataChr1 -snp gtDataChr1 -out BAll -var eigen -nths 4</div>
<div>reshuffle BAll --chi=25</div><div>[no SNP-trait pairs identified at chi2>25]</div><div><br></div><div>To make sure that actually the data are right, I ran the following tests in R making sure that the SNP is present in both big and thinned data set</div>
<div><br></div><div>> library(DatABEL)</div><div>DatABEL v.0.9-4 (March 12, 2013) loaded</div><div><br></div><div>> g <- databel("gtDataChr1")</div><div>> gs <- databel("gtDataChr1small")</div>
<div>> p <- databel("phDataChr1")</div><div>> snp <- as(g[,"rs3738547"],"vector")</div><div>> snps <- as(gs[,"rs3738547"],"vector")</div><div>> ph <- as(p[,"GI_7657133.S"],"vector")</div>
<div>> summary(lm(ph~snp))</div><div><br></div><div>Call:</div><div>lm(formula = ph ~ snp)</div><div><br></div><div>Residuals:</div><div> Min 1Q Median 3Q Max </div><div>-0.33692 -0.04485 0.00333 0.05632 0.13978 </div>
<div><br></div><div>Coefficients:</div><div> Estimate Std. Error t value Pr(>|t|) </div><div>(Intercept) 0.009116 0.005337 1.708 0.0892 . </div><div>snp -0.091529 0.015967 -5.732 3.53e-08 ***</div>
<div>---</div><div>Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1</div><div><br></div><div>Residual standard error: 0.07357 on 204 degrees of freedom</div><div>Multiple R-squared: 0.1387,<span class="Apple-tab-span" style="white-space:pre"> </span>Adjusted R-squared: 0.1345 </div>
<div>F-statistic: 32.86 on 1 and 204 DF, p-value: 3.527e-08</div><div><br></div><div>> summary(lm(ph~snps))</div><div><br></div><div>Call:</div><div>lm(formula = ph ~ snps)</div><div><br></div><div>Residuals:</div><div>
Min 1Q Median 3Q Max </div><div>-0.33692 -0.04485 0.00333 0.05632 0.13978 </div><div><br></div><div>Coefficients:</div><div> Estimate Std. Error t value Pr(>|t|) </div><div>(Intercept) 0.009116 0.005337 1.708 0.0892 . </div>
<div>snps -0.091529 0.015967 -5.732 3.53e-08 ***</div><div>---</div><div>Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1</div><div><br></div><div>Residual standard error: 0.07357 on 204 degrees of freedom</div>
<div>Multiple R-squared: 0.1387,<span class="Apple-tab-span" style="white-space:pre"> </span>Adjusted R-squared: 0.1345 </div><div>F-statistic: 32.86 on 1 and 204 DF, p-value: 3.527e-08</div><div><br></div><div><br></div>
<div><br></div><div>And this is why I blame this on reshuffle:</div><div><br></div><div>yuryaulchenko$ perl extractCell.pl BAll 1029 115001</div><div> 0.00911574810743332 -0.0915286764502525 0.00573702435940504 0.0171629786491394 -2.74352405540412e-05</div>
<div>YAMacBookPro:analysisOfHapMapeQTL yuryaulchenko$ perl extractCell.pl BAllSmall 1029 1151</div><div> 0.00911574810743332 -0.0915286764502525 0.00573702435940504 0.0171629786491394 -2.74352405540412e-05</div><div><br></div>