Trying to recode output from imputed_sex to 0 or 1

Hail Query & hailctl
1

I am trying to run a GWAS in All of Us and want to use genetic sex. However, when I use imputed_sex, from my understanding it is a boolean of is_female where female is True and male is False. To my understanding, for a GWAS everything needs to be either 0 or 1 and I cannot figure out how to recode this to be 0 or 1 within the hail environment. Thanks for any help

2

Hi @hmseagle! You can use the annotate method to add new fields to a hail table or the transmute method to overwrite an existing field. Note that for matrix tables, there are instead annotate_rows, annotate_cols and annotate_entries methods (and the same for transmute). Here’s an example using transmute to change the representation of a field:

In [3]: t = hl.utils.range_table(10)
   ...: t = t.annotate(sex=t.idx % 2 == 0)
   ...: t.show()
+-------+-------+
|   idx |   sex |
+-------+-------+
| int32 |  bool |
+-------+-------+
|     0 |  True |
|     1 | False |
|     2 |  True |
|     3 | False |
|     4 |  True |
|     5 | False |
|     6 |  True |
|     7 | False |
|     8 |  True |
|     9 | False |
+-------+-------+

In [4]: t = t.transmute(sex=hl.int(t.sex))
   ...: t.show()
+-------+-------+
|   idx |   sex |
+-------+-------+
| int32 | int32 |
+-------+-------+
|     0 |     1 |
|     1 |     0 |
|     2 |     1 |
|     3 |     0 |
|     4 |     1 |
|     5 |     0 |
|     6 |     1 |
|     7 |     0 |
|     8 |     1 |
|     9 |     0 |
+-------+-------+
3

@danielgoldstein Thanks so much for this! Would you mind if I ask you a specific question with my code? I am new to coding and especially Python. This is what I have done so far:

mt = mt.annotate_cols(pheno = phenotypes[mt.s]) #this is my matrix table that has phenotype and genotype information

imputed_sex = hl.impute_sex(mt.GT)

mt = mt.annotate_cols(imputed_sex = imputed_sex[mt.s])

In this step below, I am trying to convert the ‘True’ in the is_female section of the imputed_sex structure to 0 and “False” to 1. From my understanding, Python automatically changes True to 1 and False to 0, but the ~ will flip this, which is what I need. However, I get an error saying “MatrixTable instance has no field, method, or property ‘transmute’” I am not sure which function would be appropriate for a specific part of a structure. Thanks so much! :

mt = mt.transmute(imputed_sex=hl.int(~mt.imputed_sex.is_female))

4

Good question. In this case, there is no transmute method because you need to specify the “axis” that you are operating on (while for Tables there is only one axis, the vector of rows). Since you created imputed_sex with mt.annotate_cols, it is what we refer to as a “column field”. In order to transmute that field you need to do mt.transmute_cols. I believe what you have there should work.

You can also avoid transmute_cols by doing the negation and cast to int inside the annotate_cols, but that is a more subjective matter of readability. The output and performance should be identical.