Kindepth for double marriage

[LouisLeNezet]

Hi @sinnweja,

I'm nearly done for the new package.
I've recently discovered a subtle use case where the kindepth is not nicely computed:

Here is the code:

df <- data.frame(
    id = 1:12,
    dadid = c(0, 0, 1, 0, 0, 0, 3, 3, 5, 5, 7, 10),
    momid = c(0, 0, 2, 0, 0, 0, 4, 4, 6, 6, 9, 8),
    sex = c(1, 2, 1, 2, 1, 2, 1, 2, 2, 1, 1, 2)
)
pedi <- with(df, pedigree(id, dadid, momid, sex))
kindepth(pedi, align = TRUE)
# 0 0 1 1 0 0 2 2 1 1 3 3

We get

What would be better is to to have a kindepth of

kindepth(pedi, align = TRUE)
# 0 0 1 1 1 1 2 2 2 2 3 3

The problem seems to come in the loop that update the agood ancestors.
In the case of a double marriage like in the previous pedigree, the loop will iterate and will add the individuals as follow:

# Initial depth
0 0 1 0 0 0 2 2 1 1 3 3

# id 4 single marryin corrected
0 0 1 1 0 0 2 2 1 1 3 3

# Selected pairs is 9, 7
# abad is
9 5 6
# agood is
7 1 2 3 4

# Then by iteration agood become
# + spouse
7 1 2 3 4
# + ancestors
7 1 2 3 4
# + kids
7 1 2 3 4 8
# + spouse
7  1  2  3  4  8 10
# + ancestors
7  1  2  3  4  8 10  5  6
# + kids
7  1  2  3  4  8 10  5  6  9

The individuals 10, 5, 6 and 9 should not be added to be able to modify them.
What would be the best way to control for this case ?

Best,
Louis

[LouisLeNezet]

What I've found to control that is to do the following:

if (length(abad) == 1 && sum(c(dads, moms) == bad) == 1) {
...
else {
    ## Ancestors of the 'good' side
    agood <- c(good, ancestors(good, midx, didx))
    ## If individual already in agood they aren't really bad
    abad1 <- abad[!abad %in% agood]
    ## For spouse chasing, I need to exclude the given pair
    tdad <- dads[-who]
    tmom <- moms[-who]
    ## Get all individuals affiliated to agood
    while (TRUE) { ... }
    ## Update agood but only if not in abad1
    agood <- agood[!agood %in% abad1]
    ...
}

By doing so we keep a good kindepth measure in all case I have for now.
What do you think ?

[sinnweja]

Louis, If your solution fixes the problem with aligning subjects 5,6,9,10 with the correct generation, while also not breaking other test cases, I am in favor. Are you making the fix in a branch of kinship2, or in your Bioconductor version, or both?

…

-Jason From: Louis LE NEZET ***@***.***> Sent: Tuesday, November 14, 2023 7:17 AM To: mayoverse/kinship2 ***@***.***> Cc: Sinnwell, Jason P. ***@***.***>; Mention ***@***.***> Subject: [EXTERNAL] Re: [mayoverse/kinship2] Kindepth for double marriage (Issue #33) What I've found to control that is to do the following: if (length(abad) == 1 && sum(c(dads, moms) == bad) == 1) { ... else { ## Ancestors of the 'good' side agood <- c(good, ancestors(good, midx, didx)) ## If individual already in agood they aren't really bad abad1 <- abad[!abad %in% agood] ## For spouse chasing, I need to exclude the given pair tdad <- dads[-who] tmom <- moms[-who] ## Get all individuals affiliated to agood while (TRUE) { ... } ## Update agood but only if not in abad1 agood <- agood[!agood %in% abad1] ... } By doing so we keep a good kindepth measure in all case I have for now. What do you think ? - Reply to this email directly, view it on GitHub<#33 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AB574QERHUVAUSPXSW7YSW3YENVOBAVCNFSM6AAAAAA7KU3HG6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQMJQGE4DOMBZGY>. You are receiving this because you were mentioned.Message ID: ***@***.******@***.***>>

[LouisLeNezet]

Hi Jason,

I've added this change to the current "Pedixplorer" package for Bioconductor.
But I can do a PR to kinship2 with this change, if you want.