I ran PhaseByTransmission and the output is just unphased genotype (except when mother's genotype is missing and newborn homozygotes or the opposite). I used MendelianViolationsFile option and here is the head of the output file :
CHROM POS AC FAMILY TP MOTHER_GT MOTHER_DP MOTHER_AD MOTHER_PL FATHER_GT FATHER_DP FATHER_AD FATHER_PL CHILD_GT CHILD_DP CHILD_AD CHILD_PL
1 753541 null CC1 null G/A
.:. . . . . G/A -1 . .
1 753541 null CC10 null G/G.:. . . . . G/G -1 . .
1 753541 null CC100 null G/A.:. . . . . G/A -1 . .
1 753541 null CC102 null G/G.:. . . . . G/G -1 . .
1 753541 null CC103 null G/A.:. . . . . G/A -1 . .
1 753541 null CC104 null G/G.:. . . . . G/G -1 . .
1 753541 null CC105 null G/A.:. . . . . G/G -1 . .
1 753541 null CC106 null G/G.:. . . . . G/G -1 . .
1 753541 null CC107 null G/G.:. . . . . G/G -1 . .
Why are they mendelian violations? Is there a way to ignore the violations?
I read from Laurent's answer (http://gatkforums.broadinstitute.org/gatk/discussion/3501/unified-genotyper-gt-does-not-match-pl-causes-problems-with-phasebytransmission) that PhaseByTransmission is just using PLs and it is missing here, could it be the issue?