About the letter "Comments on the article, "Software for Y-Haplogroup Predictions, a Word of Caution"

MUZZIO M; RAMALLO V; MOTTI JM; SANTOS MR; LOPEZ CAMELO JS; BAILLIET, G

INTERNATIONAL JOURNAL OF LEGAL MEDICINE.

SPRINGER

Año: 2010 p. 162 - 164

0937-9827

About the letter "Comments on the article, "Software for Y-Haplogroup Predictions, a Word of Caution"DearSirs:TheletterbyDr.Atheyentitled?Commentsonthearticle,?SoftwareforY-HaplogroupPredictions,aWordofCaution?discussesapaperofourauthoringinthepresentJournal[1].Thoughatfirstglancehisopinionmayseemreasonable,acarefulanalysisrevealsaseriesofmistakescommittedbyDr.Athey.Inthementionedletter,oneofthemainconcernsofDr.Atheyistheamountofshorttandemrepeats(STRs)thatwereanalyzed.Hecommentsthatsevenmarkersshouldbeavoided,explainingthatanincreaseofthatamountaugmentstheprobabilityofassignment,statingthat?Withtheadditionofasufficientnumberofmarkers,thepredictionprobabilityforthecorrecthaplogroupcanbe?driven?past99%innearlyallcases,andthisalmostalwaysoccursbythepointwhere20markershavebeenused.?Nonetheless,hedoesnotmakereferencetoanysortofvalidationstudy,sosaid99%mustonlyrefertotheprobabilityofassignmenthisownsoftwareprovides.Whatismore,Dr.Atheyseemsobliviousthatourpaperexaminedthissoftware?spredictivevaluesindifferentcutoffpoints,upto95%ofprobabilityofassignment,stillobtaininginadequatepredictivevaluesonsaidcutoffpoint.Twelveofourhaplotypesproposedfrom100%to99.6%probabilityofassignmenttotheR1bhaplogroupintheHaplogroupPredictor,whilenoneofthosesamplesbelongtothesaidhaplogroup.IntheRhaplogroup,onehaplotypegives99.8%probabilitytoanerredhaplogroup(E1b1b),while14givefrom100%to99.4%,assigningthesesamplestotheR1bhaplogroup,andwereconsideredascorrectpredictions.Notethatinthesecases,thepredictedhaplogroupfollowsthenomenclatureofthesoftware,andtheprobabilitywasnotpooledbymajorbranches.Ifthesevenmarkersoftheminimalhaplotypewerenotaproperset,Dr.Athey?ssoftwareshouldnotprovidesuchhighprobabilitiesofassignmentinthesecases,sinceitmisleadstheuser.Unfortunately,Dr.Athey?smentionthatsevenY-STRsaretoofewoccursonlyinhisletter;neitherofthetwopapersofhisauthoring[2,3]availableattheHaplogroupPredictor?swebsite(http://www.hprg.com/hapest5/)northesoftwareinstructions(http://www.hprg.com/hapest5/page4.html)explainthattheusershouldemploymorethansevenmarkers.Otherwise,wewouldnothaveattemptedtouseit.Moreover,asimpleglanceatrecentliteraturewheretheHaplogroupPredictorwasemployedclearlyshowsthatresearchersdonotusesuchhighamountsofY-STRswhentheyrelyontheHaplogroupPredictor;forinstance,Salasetal.[4]usedthesevenY-STRswedid,plusDYS385andPetrejcikovaetal.[5]only12Y-STRs.