Skip to main content

Wellstone bilateral tibialis anterior FSHD biopsied muscle study

11 DNA methylation of FSHD D4Z4 allele

We conducted bisulfite sequencing on bilateral TA muscle biopsies from 34 subjects. In this chapter, we present the following:

  • The CpG methylation percentages for subjects with the 4qA161S and 4qA161L alleles.
  • The correlation of methylation levels between left and right TA biopsies.
  • The association between methylation and STIR status.
  • Correlations with other variables, including muscle strength, disease molecular signatures, whole muscle fat percentage, and clinical severity score (CSS).

Note: We excluded subject 01-0022 in this analysis due to its duplication in D4Z4 repeat regions.

11.1 Loading and cleaning data

We load two data sets from the data directory: 1. comprehensive_df.rda: clinical and MRI characteristic data 2. bs_methyl.rda: DNA methylation data

library(tidyverse)
library(corrr)
library(kableExtra)

pkg_dir <- "/Users/cwon2/CompBio/Wellstone_BiLateral_Biopsy"
load(file.path(pkg_dir, "data", "bs_methyl.rda"))
load(file.path(pkg_dir, "data", "comprehensive_df.rda"))
comprehensive_df <- comprehensive_df %>%
  dplyr::rename_with(~str_remove(., '-logSum'), ends_with('-logSum'))

Tidy up the bisulfite sequencing data.

bss <- bs_methyl %>% 
  dplyr::mutate(`4A161S` = if_else(grepl("4A161S", `haplotype_chr4`),
                                   TRUE, FALSE)) %>% 
  dplyr::mutate(`4A161L` = if_else(grepl("4A161L", `haplotype_chr4`),
                                   TRUE, FALSE)) %>%
  dplyr::filter(! Subject == "01-0022") %>%
  dplyr::mutate(allele=if_else(`use BSSL`, "4qA161L", "4qA161S")) %>%
  dplyr::mutate(allele=factor(allele, levels=c("4qA161S", "4qA161L")))
Table 11.1: Tidy table of CpG methylation percentage and allele.
Subject location sample_id allele BSS Value Predicted Haplotype
01-0019 R 01-0019R 4qA161S 8.000000 4A161S(5RU)/4B163(32RU), 10A164/10A164
01-0019 L 01-0019L 4qA161S 4.000000 4A161S(5RU)/4B163(32RU), 10A164/10A164
01-0020 R 01-0020R 4qA161S 4.000000 4A161S(5RU)/4A161S(22RU), 10A166/10A166
01-0020 L 01-0020L 4qA161S 4.000000 4A161S(5RU)/4A161S(22RU), 10A166/10A166
01-0021 R 01-0021R 4qA161L 14.285710 4A161L(10RU)/4B163(16RU), 10A166/10A166
01-0021 L 01-0021L 4qA161L 19.047620 4A161L(10RU)/4B163(16RU), 10A166/10A166
01-0023 R 01-0023R 4qA161S 8.000000 4A161S/4B163, 10A166/10A166
01-0023 L 01-0023L 4qA161S 8.000000 4A161S/4B163, 10A166/10A166
01-0024 R 01-0024R 4qA161S 8.000000 4A161S(7RU)/4B163(17RU), 10A166/10A166
01-0024 L 01-0024L 4qA161S 4.000000 4A161S(7RU)/4B163(17RU), 10A166/10A166
01-0025 R 01-0025R 4qA161S 4.000000 4A161S(6RU)/4A161S(16RU), 10A166/10A166
01-0025 L 01-0025L 4qA161S 4.000000 4A161S(6RU)/4A161S(16RU), 10A166/10A166
01-0026 R 01-0026R 4qA161S 0.000000 4A161S(6RU)/4A161L(45RU), 10A166/10A168
01-0026 L 01-0026L 4qA161S 0.000000 4A161S(6RU)/4A161L(45RU), 10A166/10A168
01-0028 R 01-0028R 4qA161S 8.000000 4A161S(7RU)/4A161S(33RU), 10A166/4A166
01-0028 L 01-0028L 4qA161S 8.000000 4A161S(7RU)/4A161S(33RU), 10A166/4A166
01-0029 R 01-0029R 4qA161S 4.000000 4A161S(7RU?)/4A161S, 10A166/4A166
01-0029 L 01-0029L 4qA161S 0.000000 4A161S(7RU?)/4A161S, 10A166/4A166
01-0030 R 01-0030R 4qA161S 4.166667 4A161S(5RU)/4B168(31RU), 10A166/4A166
01-0030 L 01-0030L 4qA161S 12.000000 4A161S(5RU)/4B168(31RU), 10A166/4A166
01-0031 R 01-0031R 4qA161S 12.000000 4A161S(7RU)/4B163(34RU), 10A166/10A166
01-0031 L 01-0031L 4qA161S 8.000000 4A161S(7RU)/4B163(34RU), 10A166/10A166
13-0001 R 13-0001R 4qA161S 4.000000 4A161S(4RU)/4A161L(16RU:52RU=1:1), 10A166/10B161T
13-0001 L 13-0001L 4qA161S 0.000000 4A161S(4RU)/4A161L(16RU:52RU=1:1), 10A166/10B161T
13-0002 R 13-0002R 4qA161S 4.000000 4A161S(5RU)/4B163(28RU), 10A166/10A166
13-0002 L 13-0002L 4qA161S 4.000000 4A161S(5RU)/4B163(28RU), 10A166/10A166
13-0003 R 13-0003R 4qA161S 12.000000 4A161S/4B163, 10A166/10A166
13-0003 L 13-0003L 4qA161S 8.000000 4A161S/4B163, 10A166/10A166
13-0004 R 13-0004R 4qA161S 4.000000 4A161S(4RU)/4A166H(27RU), 10A166/10A166
13-0004 L 13-0004L 4qA161S 8.000000 4A161S(4RU)/4A166H(27RU), 10A166/10A166
13-0005 R 13-0005R 4qA161S 8.000000 4A161S(9RU)/4B163(23RU), 10A166/10A166
13-0005 L 13-0005L 4qA161S 16.000000 4A161S(9RU)/4B163(23RU), 10A166/10A166
13-0006 L 13-0006L 4qA161L 8.000000 4A161/4A161L, 10A166/10A166
13-0007 R 13-0007R 4qA161L 42.857140 4A161L(2RU)/4B163(27RU), 10A164/10A166
13-0007 L 13-0007L 4qA161L 38.095240 4A161L(2RU)/4B163(27RU), 10A164/10A166
13-0008 R 13-0008R 4qA161S 4.000000 4A161S/4A161S, 10A166/10A166
13-0008 L 13-0008L 4qA161S 8.333333 4A161S/4A161S, 10A166/10A166
13-0009 R 13-0009R 4qA161S 4.166667 4A161S(5RU)/4A161L(22RU), 10A166/10A166
13-0009 L 13-0009L 4qA161S 4.000000 4A161S(5RU)/4A161L(22RU), 10A166/10A166
13-0010 R 13-0010R 4qA161S 4.000000 4A161S(8RU)/4A161S(21RU), 10A164/10A166
13-0010 L 13-0010L 4qA161S 8.000000 4A161S(8RU)/4A161S(21RU), 10A164/10A166
13-0011 R 13-0011R 4qA161S 12.000000 4A161S(6RU)/4B168(24RU), 10A166/10A166
13-0011 L 13-0011L 4qA161S 24.000000 4A161S(6RU)/4B168(24RU), 10A166/10A166
32-0020 R 32-0020R 4qA161L 9.523810 4A161L(6RU?)/4B168, 10A166/10A166
32-0020 L 32-0020L 4qA161L 4.761905 4A161L(6RU?)/4B168, 10A166/10A166
32-0021 R 32-0021R 4qA161S 8.000000 4A161S(3RU)/4B168(117RU), 10A166/10A166
32-0021 L 32-0021L 4qA161S 8.000000 4A161S(3RU)/4B168(117RU), 10A166/10A166
32-0022 R 32-0022R 4qA161S 4.000000 4A161S(7RU)/4A168(46RU), 10A166/10A166
32-0022 L 32-0022L 4qA161S 8.000000 4A161S(7RU)/4A168(46RU), 10A166/10A166
32-0023 R 32-0023R 4qA161S 8.000000 4A161S(9RU)/4B163(25RU), 10A166/10A166
32-0023 L 32-0023L 4qA161S 12.000000 4A161S(9RU)/4B163(25RU), 10A166/10A166
32-0024 R 32-0024R 4qA161S 8.000000 4A161S(9RU)/4A161S(36RU), 10A164/10A166
32-0024 L 32-0024L 4qA161S 12.000000 4A161S(9RU)/4A161S(36RU), 10A164/10A166
32-0025 R 32-0025R 4qA161S 12.000000 4A168S(4RU)/4B168(42RU), 10A164/10A166
32-0025 L 32-0025L 4qA161S 8.000000 4A168S(4RU)/4B168(42RU), 10A164/10A166
32-0026 R 32-0026R 4qA161L 4.761905 4A161L(6RU)/4B168(16RU), 10A166/10A166
32-0026 L 32-0026L 4qA161L 4.761905 4A161L(6RU)/4B168(16RU), 10A166/10A166
32-0027 R 32-0027R 4qA161S 16.000000 4A161S/4A166, 10A164/10A166
32-0027 L 32-0027L 4qA161S 20.000000 4A161S/4A166, 10A164/10A166
32-0028 R 32-0028R 4qA161S 5.125000 4A161S(5RU)/4A166H(10RU), 10A166/10A166
32-0028 L 32-0028L 4qA161S 4.000000 4A161S(5RU)/4A166H(10RU), 10A166/10A166
32-0029 R 32-0029R 4qA161S 0.000000 4A161S(5RU)/4A166H(40RU), 10A164/10A166
32-0029 L 32-0029L 4qA161S 4.000000 4A161S(5RU)/4A166H(40RU), 10A164/10A166
32-0030 R 32-0030R 4qA161S 0.000000 4A161S(4RU)/4B163(33RU), 10A166/10A166
32-0030 L 32-0030L 4qA161S 4.000000 4A161S(4RU)/4B163(33RU), 10A166/10A166

11.2 4qA161S and 4qA161L allele

Mean values of 4qA161S and 4qA161L allele:

bss %>%
  group_by(`use BSSL`) %>%
  summarise(mean=mean(`BSS Value`),  min=min(`BSS Value`), max=max(`BSS Value`))
#> # A tibble: 2 × 4
#>   `use BSSL`  mean   min   max
#>   <lgl>      <dbl> <dbl> <dbl>
#> 1 FALSE       6.96  0     24  
#> 2 TRUE       16.2   4.76  42.9

bss %>% dplyr::filter(!Subject=="01-0022") %>%
  dplyr::mutate(allele=if_else(`use BSSL`, "4qA161L", "4qA161S")) %>%
  dplyr::mutate(allele=factor(allele, levels=c("4qA161S", "4qA161L"))) %>%
  ggplot(aes(x=allele, y=`BSS Value`)) +
    geom_boxplot(width=0.5, outlier.shape=NA) +
    geom_jitter(width=0.2, alpha=0.3, size=1, color="grey50") +
    theme_minimal() +
    labs(x="", y="% methylation") +
    theme(axis.text.x = element_text(angle = 25, vjust = 1, hjust=1))
The CpG methylation percentage of subjects with 4qA161S and 4qA161L allele.

Figure 11.1: The CpG methylation percentage of subjects with 4qA161S and 4qA161L allele.

11.3 Left and right TA biopsies correlation 

bss_cor <- bss %>% dplyr::select(Subject, location, `BSS Value`) %>%
  spread(key=location, value=`BSS Value`) %>%
  dplyr::select(-Subject) %>%
  corrr::correlate() %>% dplyr::filter(!is.na(L)) %>% pull(L) 

bss %>% dplyr::select(Subject, location, `BSS Value`) %>%
  spread(key=location, value=`BSS Value`) %>%
  ggplot(aes(x=L, y=R)) +
    geom_point() +
    geom_smooth(method="lm", se=FALSE, linewidth=0.7, color="grey75", alpha=0.3) +
    annotate("text", x=Inf, y=-Inf, color="gray25",
             size=3, hjust=1, vjust=-2,
             label=paste0("Pearson=", format(bss_cor, digit=2))) +
    theme_minimal() +
    labs(x="L (% methylation)", y="R (% methylation)")
#> Warning: Removed 1 rows containing non-finite values
#> (`stat_smooth()`).
#> Warning: Removed 1 rows containing missing values
#> (`geom_point()`).
Methylation correlation between left and right TA biopsies: Pearson correlation = 0.84.

Figure 11.2: Methylation correlation between left and right TA biopsies: Pearson correlation = 0.84. 

#> Warning: Removed 1 rows containing non-finite values
#> (`stat_smooth()`).
#> Warning: Removed 1 rows containing missing values
#> (`geom_point()`).

11.4 Methylation and STIR status 

bss %>% dplyr::select(-Subject) %>%
  left_join(comprehensive_df, by="sample_id") %>% 
  ggplot(aes(x=STIR_status, y=`BSS Value`)) +
    geom_boxplot(width=0.5, outlier.shape=NA) +
    geom_jitter(width=0.3, alpha=0.3, size=1, color="grey50") +
    facet_wrap(~allele) +
    theme_bw() +
    labs(x="", y="% methylation")
The methylation levels are not associated with STIR status.

Figure 11.3: The methylation levels are not associated with STIR status.

11.5 Methylation and other variables

In the figure below we demonstrated that there is no association between CpG methylation and other variables—whole muscle fat, CSS, and muscle strength—and mRNA levels of disease signatures.

cor_161S <- bss %>% dplyr::select(-Subject) %>%
  dplyr::filter(`allele`=="4qA161S") %>%
  left_join(comprehensive_df, by="sample_id") %>% 
  dplyr::select(`BSS Value`, `DUX4-M6`, `Foot Dorsiflexors`,
                `CSS`, `Fat_Infilt_Percent`, `Cumulative Score`) %>%
  corrr::correlate() %>% 
  select(term, `BSS Value`)
cor_161L <- bss %>% dplyr::select(-Subject) %>%
  dplyr::filter(`allele`=="4qA161L") %>%
  left_join(comprehensive_df, by="sample_id") %>% 
  dplyr::select(`BSS Value`, `DUX4-M6`, `Foot Dorsiflexors`,
                `CSS`, `Fat_Infilt_Percent`, `Cumulative Score`) %>%
  corrr::correlate() %>% 
  select(term, `BSS Value`) 

cor_161S %>% 
  full_join(cor_161L, by="term", suffix=c("4qA161S", "4qA161L")) %>%
  drop_na() %>%
  rename(`4qA161S`=`BSS Value4qA161S`, `4qA161L`=`BSS Value4qA161L`) %>%
  gather(key=allele, value=cor, -term) %>%
  ggplot(aes(x=allele, y=term)) +
    geom_tile(aes(fill=cor), colour = "grey50") +
    geom_text(aes(label=format(cor, digit=1)), size=3) +
    scale_fill_steps2(low="grey75", mid="grey95", high="grey75", 
                     breaks=seq(-1, 1, length.out=10)) +
    theme_minimal() +
    labs(x="", y="", title="Pearson correlation") +
    theme(legend.position = "none", 
          panel.grid.major = element_blank(),
          panel.grid.minor = element_blank())
Pearson correlation between methylation levels and other variables including mRNA levels of disease signatures, whole muscle fat, CSS, and muscle strengh.

Figure 11.4: Pearson correlation between methylation levels and other variables including mRNA levels of disease signatures, whole muscle fat, CSS, and muscle strengh.