Mutation in cancers

Recurrent mutation in cancers

• The core idea is to study recurrent SNP mutations in cancers:
  • their distribution
    
  • their effects
    • suppress
    • enhance
    • change function?
  • why they are recurrent,possibly:
    • DNA level
      
    • Protein level
      
  • usage: ??

The goal: to answer these questions:

1. What’s the distribution of recurrent mutations in cancer / non-cancer genes?

   Recurrent mutations are enriched in both oncogenes and tumor suppressors.

	1<#<5	>5	percent_gt5	enrich
NON_CG	261137	2999	0.011	1.000
POG	5817	205	0.034	2.998
TSG	6891	180	0.025	2.242

2. What’s the distribution of SNPs in POG/TSG

3. For recurrent nonsense mutations, where did they cut, what are their effects?

   Case study of p53

   In short: recurrent nonsense mutations might due to DNA methylation sites.

require(ggplot2)
require(dplyr)
require(ggrepel)
source("../R/seq_potentialns.R")
source("../R/plot_pfamdomain.R")
source("../R/pfam_readpfam.R")
load("../test2/tp53.cdna.rda")
dist<-seq_potentialns(tp53.cdna)
load("../data/tp53test2.rda")
ungroup(minimaf.tp53.sum)->minimaf.tp53.sum
nshotspot<-minimaf.tp53.sum %>%
  filter(Variant_Classification=="Nonsense_Mutation") %>% 
  group_by(Protein_Change) %>% 
  summarise(recur_allcoh=sum(recur_pos)) %>% 
  mutate(Prot_pos=as.numeric(gsub("[^0-9]|_.*", "", Protein_Change))) %>%
  filter(recur_allcoh>1)
gg <-
  ggplot(data = dist %>% filter(dist == 1)) +
  ## potential site
  geom_point(aes(x = x, y = dist),
  size = 3,
  alpha = 0.2) +
  ## lollipop 
  geom_segment(
  inherit.aes = FALSE,
  data = nshotspot %>% filter(recur_allcoh > 5),
  aes(
  x = Prot_pos,
  y = recur_allcoh,
  xend = Prot_pos,
  yend = 0
  )
  ) +
  geom_point(
  inherit.aes = FALSE,
  data = nshotspot %>% filter(recur_allcoh > 5),
  aes(x = Prot_pos, y = recur_allcoh),
  colour = "red",
  size = 3
  ) +
  
  ggrepel::geom_text_repel(
  inherit.aes = FALSE,
  data = nshotspot %>% filter(recur_allcoh > 19),
  aes(x = Prot_pos, y = recur_allcoh, label = Protein_Change),
  colour = "blue"
  
  ) +
  plot_pfamdomain("P04637") +
  labs(fill = "Domain", y = "Recurrence", x = "AA position")
  
  plot(gg)

Nonsense mutations reduce protein expression level at mRNA level:

In Breast cancer (BRCA), gene :TP53(tsg), CDH1(tsg), MAP3K1(unclear)

4. For recurrent silent mutations, will they affect expression?

   Will do this with 5

5. What are the effects of recurrent mutations to neighboring genes

   Working on
   • plan:

     1. retrieve recurrent mutated gene and its neighbor in network

     2. retrieve expression profile in a cancer cohort for this gene set

        • patient level

     3. Comparison & statistical test

   

6. Mutation relations: co-mutated / mutation exclusive

   • Other studies: gene level.

Available data

• TCGA Data: CNA, Mutation, Expression(protein/mRNA), clinical output
  

• COSMIC data : Mainly mutations.

• Additional
  • pathway commons : to identify neighbors
    
  • pfam : protein domain info
    
  • cancer gene: Cancer gene census and uniprot POG/TSG keyword