Hi Pasi!
I used ParentCall2 to convert a .vcf.gz file to data.call.gz, but I noticed that most of the values in data.call.gz are 1. I'm not sure what the problem is. The .vcf file was generated from chip genotypes using plink, so it only contains FMT/GT information and does not include likelihood values.
The most important thing is that I only got two LGs by SeparateChromosomes2（lod=10 theta=0.08）; but my family has 1483；
Are you sure that the individual names match in your pedigree and vcf? If not, ParentCall2 will give a warning that the individual(s) are not found and are set to all missing... This would explain all 1s (missing) in your data.
ParentCall2 should read GT field if it does not find PL or GL.
Cheers,
Pasi
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Dear pasi,
The VCF file contains 6715 samples, with over 4000 individuals whose sample IDs can be found in pedigree.txt. There are more than 1000 families, each of which includes parents who have at least one genotype and their offsprings also have genotypes. I am unsure if it is problematic for only the father or mother to be genotyped in the parents' case.
In the screenshot I provided above, the familyID is LL14CLA17701435_LL22RK917700771. Among them, LL22RK917700771 has a genotype in the vcf file, while LL14CLA17701435 does not (i.e., it has no genotype). Their offspring LLGXRXC18008002 has a genotype. However, after transformation, all individuals in the family and SNPs appear to be displayed as 1.0.
thanks
chinj cheung
Last edit: chinj cheung 2024-03-07
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
If you have a lot of small families with only one parent, it might happen that the parental genotypes cannot be inferred. In this case you also obtain all 1s. You can try to lower familyLimit parameter (try 1 or 1.5) if this is the case. For a single parent family, you might need 5-8 offspring to call the missing parental genotype. If you have halfsib families, they might help. Also the parameter ignoreParentOrder in ParentCall2 can be tried but that makes the map construction more difficult (it requires extra parameters in other modules as well).
Cheers,
Pasi
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Dear pasi,
Yes! I have approximately 1200 single-parent families and around 200 two-parent families. To clarify, when running ParentCall2, should I add the parameters familyLimit=1(1.5) and ignoreParentOrder=1, where ignoreParentOrder means that the order of the parents' genotypes need not be taken into account?
But this statement "makes the map construction more difficult (it requires extra parameters in other modules as well)" implies that I should pay attention to what during the subsequent map construction?
thanks
chinj cheung
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Lep-MAP3 requires data with both parental genotypes. ParentCall2 tries to impute missing parents if it is possible. This imputation is difficult if you have only few offspring and no grandparents or half-sib families.
Family limit defines how certain the imputed parental genotypes must be. Lowering it might yield more data but there might be more errors as well.
I have constructed good maps with ignoreParentOrder as well. This might help more if both parents are missing as it keeps families where it can be detected that exactly one parent is informative but it is not known which. However, with small families without both parental genotypes, it is difficult to say whether both parents are informative or only one.
Cheers,
Pasi
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi Pasi!
I used ParentCall2 to convert a .vcf.gz file to data.call.gz, but I noticed that most of the values in data.call.gz are 1. I'm not sure what the problem is. The .vcf file was generated from chip genotypes using plink, so it only contains FMT/GT information and does not include likelihood values.
The most important thing is that I only got two LGs by SeparateChromosomes2（lod=10 theta=0.08）; but my family has 1483；
Thanks,
chinj cheung
When lod=1, the values of theta are 0.08, 0.02, and 0.04. At this time, there are still only two LGs.
Dear chinj cheung,
Are you sure that the individual names match in your pedigree and vcf? If not, ParentCall2 will give a warning that the individual(s) are not found and are set to all missing... This would explain all 1s (missing) in your data.
ParentCall2 should read GT field if it does not find PL or GL.
Cheers,
Pasi
Dear pasi,
The VCF file contains 6715 samples, with over 4000 individuals whose sample IDs can be found in pedigree.txt. There are more than 1000 families, each of which includes parents who have at least one genotype and their offsprings also have genotypes. I am unsure if it is problematic for only the father or mother to be genotyped in the parents' case.
In the screenshot I provided above, the familyID is LL14CLA17701435_LL22RK917700771. Among them, LL22RK917700771 has a genotype in the vcf file, while LL14CLA17701435 does not (i.e., it has no genotype). Their offspring LLGXRXC18008002 has a genotype. However, after transformation, all individuals in the family and SNPs appear to be displayed as 1.0.
thanks
chinj cheung
Last edit: chinj cheung 2024-03-07
Dear chinj cheung,
Wow, this is a huge dataset!
If you have a lot of small families with only one parent, it might happen that the parental genotypes cannot be inferred. In this case you also obtain all 1s. You can try to lower familyLimit parameter (try 1 or 1.5) if this is the case. For a single parent family, you might need 5-8 offspring to call the missing parental genotype. If you have halfsib families, they might help. Also the parameter ignoreParentOrder in ParentCall2 can be tried but that makes the map construction more difficult (it requires extra parameters in other modules as well).
Cheers,
Pasi
Dear pasi,
Yes! I have approximately 1200 single-parent families and around 200 two-parent families. To clarify, when running ParentCall2, should I add the parameters familyLimit=1(1.5) and ignoreParentOrder=1, where ignoreParentOrder means that the order of the parents' genotypes need not be taken into account?
But this statement "makes the map construction more difficult (it requires extra parameters in other modules as well)" implies that I should pay attention to what during the subsequent map construction?
thanks
chinj cheung
Dear chinj cheung,
Lep-MAP3 requires data with both parental genotypes. ParentCall2 tries to impute missing parents if it is possible. This imputation is difficult if you have only few offspring and no grandparents or half-sib families.
Family limit defines how certain the imputed parental genotypes must be. Lowering it might yield more data but there might be more errors as well.
I have constructed good maps with ignoreParentOrder as well. This might help more if both parents are missing as it keeps families where it can be detected that exactly one parent is informative but it is not known which. However, with small families without both parental genotypes, it is difficult to say whether both parents are informative or only one.
Cheers,
Pasi
Dear pasi,
Thank you for your reply.
thanks,
Chinj Cheung