Closed sadamowi closed 7 years ago
Hi Sally, I do have a small component of code that eliminates pairings with relative outgroup distance of 1 (no difference in outgroup distance) at line 1033 which I posted below. Is this what you meant?
equalOutGroupDistCheck <- foreach(i=1:nrow(dfPairingResultsL1)) %do% (dfPairingResultsL1$outGroupDistance[i] - dfPairingResultsL2$outGroupDistance[i]) equalOutGroupDistCheck <- which(equalOutGroupDistCheck == 0)
if(length(equalOutGroupDistCheck)>0){
dfPairingResultsL1 <- dfPairingResultsL1[-equalOutGroupDistCheck,]
dfPairingResultsL2 <- dfPairingResultsL2[-equalOutGroupDistCheck,] }
Thank you for explaining this part. I missed that. Issue closed.
Hi Matt,
I am wondering if the current pipeline is biasing the results a little bit towards "failures". I notice there is code for designating "successes" (defined as a directional result). Then, "failures" seem to be not successes.
Line 1261:
Also making a vector for pairings that were not successes
failureOverall <- dfPairingResultsL1$inGroupPairing[-successOverall]
What about cases if there is absolutely no difference in the relative outgroup distance? I suggest those cases should be omitted priot to statistical testing rather than counted as failures. Are those accounted for somewhere? Perhaps that is addressed elsewhere, and I just overlooked that.
Thanks very much for checking.
Cheers, Sally