4. Check sex discordance (Sample-level)

Sample mix-up is investigated by looking at sex mismatch, where sex information of each sample is compared with genotypes at chromosomes X and Y.

Note

NB: This method however does not detect same sex sample mix-ups and is not reliable when there are sex chromosome abnormalities.

Note

NB: Fot this step check that ped file has recorded sex, otherwise consider using sex from metadata

For this step we need to use files with already excluded low call rate SNPs

runPLINK("--ped file.ped --map file.map --mind 0.1 --recode --out file_SNP_filtered")

Now let’s check check sex discordance

runPLINK("--ped file_SNP_filtered.ped --map file_SNP_filtered.map --check-sex --out file_SNP_filtered")
system("cat file_SNP_filtered.sexcheck | tr -s ' ' '\t' >  file_SNP_filtered_sexcheck.tabs")

Plot reported sex vs F value (X heterozygosity level).

Plink_sex_filtered <- read.delim("file_SNP_filtered_sexcheck.tabs") %>%
        select(-X)
Plink_sex_filtered <- Plink_sex_filtered %>%
  mutate(PEDSEX = if_else(PEDSEX == "2", "female", "male"))

ggplot(Plink_sex_filtered, aes(x=PEDSEX, y= `F`)) +
        geom_jitter(color = ifelse((Plink_sex_filtered$PEDSEX == "female"), "red", "blue")) +
        geom_text_repel(aes(label=ifelse(((PEDSEX == "female") & ("F" > 0.2)), IID, ""),
                                                hjust = 0.4, vjust= 0.4))+
        xlab("Reported sex")+
        ylab("Heterozygosity rate")+
        theme_bw()

Note

plink report marks problematic samples, this information should be available in file_SNP_filtered_sexcheck.tabs Advice It is also important to look at the plot and asses it visually

Check Y chr counts also if there are strange samples

runPLINK("--ped file_SNP_filtered.ped --map file_SNP_filtered.map --check-sex ycount --out file_SNP_filtered_ycount")
system("cat file_SNP_filtered_ycount.sexcheck | tr -s ' ' '\t' > file_SNP_filtered_ycount_sexcheck.tabs")
Plink_sex_filtered_ycount <- read.delim("file_SNP_filtered_ycount_sexcheck.tabs") %>%
        select(-X)

Plot y chromosome counts

Plink_sex_filtered_ycount <- Plink_sex_filtered_ycount %>%
        mutate(PEDSEX = if_else(PEDSEX == "2", "female", "male"))

ggplot(Plink_sex_filtered_ycount, aes(x=PEDSEX, y= YCOUNT )) +
        geom_text_repel(aes(label=ifelse(((PEDSEX == "male") & (YCOUNT < 3000)), IID,""),
                                        hjust= 0.4, vjust= 0.4)) +
        xlab("Reported sex") +
        ylab("Y counts") +
        theme_bw()

Note

Advice: Look at the plot and check that males and females samples are clearly separating on Y-axis