How does GATK PhaseByTransmission define the phasing of Father and Mother?

Question

[shajib.sust]

Hi, I have genotyped a trio, and then phased them using PhaseByTransmission of GATK Tools. I have then parsed the output vcf file using BCFTools query command. The output file looks like as follows:

#CHR POS     FAT    MOT  CHILD      
20      65288   G/T     ./.     ./.                                                                                                            
20      65900   A|A     A|G     A|A                                                                                                                                                                                                                       
20      66720   ./.     C/A     C/A 
20      68749   T|C     C|C     C|T                                                                                                            
20      69094   G/A     ./.     ./.

The phasing information can be interpreted as Mother|Father. Now, I understand that the Child Phasing can be done by applying Mendelian Laws. But How the phasing of Father or Mother is done here? For example here

20      68749   T|C     C|C     C|T

every person of the trio is phased; even the father (het) without even knowing grand parents of the child. As I am a new researcher in this field, your clarification will help me a lot to understand phasing.

Thank you very much,

-Ruhul

[shajib.sust]

Dear Jeremy Leipzig,

Thank you very much for your reply... I will be very happy to get some more suggestions from you. Is there anyway to phase all of the child genotypes using both the PhaseByTransmission and ReadBackedPhasing for a given family trio genotype information?

With Kind Regards,
-Ruhul

[ebrown1955]

You can do PhaseByTransmission first, then ReadBackedPhasing with the -respectPhaseInput option, which will keep the PhaseByTransmission phases.

[mcclintock]

Hi, GATK4 doesn' include PhaseByTransmission. So my question is how to do that. Thanks.

Answer 1

[Jeremy Leipzig]

Yes good point. The child's genotype is MOTHER|FATHER but there is no way GATK would know for GRANDMOTHER|GRANDFATHER for either of the parents. It's just a custom to use '|' for the whole trio if the child is phased.