---
title: "COVID-19 Data Resampling"
author: "Landon Kehr"
output: 
  flexdashboard::flex_dashboard:
    theme: cosmo
    orientation: columns
    social: ["facebook", "twitter", "linkedin"]
    source_code: embed
    vertical_layout: scroll
---

```{r setup, include=FALSE}
#2.1
#rm(list=ls())
library(tidyverse)
library(dplyr)
library(ROSE)
library(RColorBrewer)
library(Rfast)
library(ggplot2)
library(lubridate)
library(flexdashboard)
#list.files("msa_1007")
#setwd(getwd()) #set working directory to source directory
```

# Introduction

## Column {.tabset data-width=600}

### Abstract

**Addressing sample biases in genome-wide association study for SARS-CoV-2**

Monitoring adaptive changes of SARS-CoV-2 is of critical importance to mitigate its transmission. There are many researches showing the viral genomic mutations play a key role in propagation of SARS-COV-2. Genome-wide association study (GWAS) is a typical method to study the association between genotype and phenotypic measures. However, GWAS typically requires random sampling. Currently, isolates of SARS-CoV-2 tend to be sequenced in regions with advanced research capacity, and most sequences were reported during March-April 2020. These, and other sampling biases pose challenges to GWAS of SARS-CoV-2.
 In this work, several statistical and computational methods to mitigate the sampling biases of SARS-COV-2 are proposed.

```{r warning=FALSE}
tb <- read_csv("tb.csv") 

```


```{r}
dim(tb)
```



```{r}
#Filtering out countries that have fewer than 50 samples

group <- tb %>% group_by(country) %>% summarize(N=n()) %>% filter(N>=50)
#sum(group$N)
#group
cleaned <- tb %>% filter(country %in% group$country)
```

### Summary of the data

```{r}
head(cleaned)
```

## Column {.tabset data-width=600}

### Original Data Set
```{r}
#Bar graph of the final cleaned data set grouped by strain clade determined by GISAID

originalBar <- ggplot(tb,aes(x=GISAID_clade)) + 
  geom_bar()
originalBar
```

### Undersample of 20
```{r}
#Create an undersample of 20 data points from each country

set.seed(2)
undersample20 <- cleaned %>% group_by(country) %>% sample_n(20)
under20Bar <- ggplot(undersample20, aes(x=GISAID_clade)) + 
  geom_bar()
under20Bar
```

### Undersample of 40
```{r}
#Create an undersample of 40 data points from each country

set.seed(1)
undersample40 <- cleaned %>% group_by(country) %>% sample_n(40)
under40Bar <- ggplot(undersample40, aes(x=GISAID_clade)) + 
  geom_bar()
under40Bar
```

### Oversample of 200
```{r}
#Create an oversample of 200 data points from each country

set.seed(3)
oversample200 <- cleaned %>% group_by(country) %>% slice_sample(n=200,replace=TRUE)
over200Bar <- ggplot(oversample200, aes(x=GISAID_clade)) + 
  geom_bar()
over200Bar
```

### Oversample of 500
```{r}
#Create an oversample of 500 data points from each country

set.seed(4)
oversample500 <- cleaned %>% group_by(country) %>% slice_sample(n=500,replace=TRUE)
over500Bar <- ggplot(oversample500, aes(x=GISAID_clade)) + 
  geom_bar()
over500Bar
```

# Something {.hidden}

## Column {.tabset}

#By Month
##Prop Table
```{r}
#Original Data Set Proportion Table by Month

mainCladeTableMonth <- table(tb$GISAID_clade, tb$month)
#mainCladeTableMonth
mainPropTableMonth <- as.data.frame(prop.table(mainCladeTableMonth,2)) %>% 
  mutate("Clade"=Var1,
        "Month"=Var2,
         "Frequency"=Freq*100) %>%
  select(Clade,Month,Frequency)
#mainPropTableMonth
```
##Frequency Plot
```{r}
#Original Data Set Frequency Percentage Plot Month

originalDateMonth <- ggplot(mainPropTableMonth,aes(x=Month,y=Frequency,fill=Clade, width=6)) + 
  geom_col() + 
  scale_x_discrete(limits=mainPropTableMonth$Month) + 
  scale_fill_brewer(palette="Paired") + 
  theme_bw()
#originalDateMonth
```

#By Week
##Prop Table
```{r}
#Original Data Set Proportion Table by Week

mainCladeTableWeek <- table(tb$GISAID_clade, tb$week)
mainPropTableWeek <- as.data.frame(prop.table(mainCladeTableWeek,2)) %>% 
  mutate("Clade"=Var1,
         "Week"=Var2,
         "Frequency"=Freq*100) %>%
  select(Clade,Week,Frequency)
#mainPropTableWeek
```
##Frequency Plot
```{r}
#Original Data Set Frequency Percentage Plot Week

originalDateWeek <- ggplot(mainPropTableWeek,aes(x=Week,y=Frequency,fill=Clade,width=6)) + 
  geom_col() + 
  scale_x_discrete(limits=mainPropTableWeek$Week) + 
  scale_fill_brewer(palette="Paired") + 
  theme_bw()
#originalDateWeek
```

#By Continent
##Prop Table
```{r}
#Original Data Set Proportion Table by Continent

mainCladeTableLocCont <- table(tb$GISAID_clade, tb$region)
mainPropTableLocCont <- as.data.frame(prop.table(mainCladeTableLocCont,2)) %>% 
  mutate("Clade"=Var1,
         "Continent"=Var2,
         "Frequency"=Freq*100) %>%
  select(Clade,Continent,Frequency)
#mainPropTableLocCont
```

##Frequency Plot
```{r}
#Original Data Set Percentage by Continent

originalCont <- ggplot(mainPropTableLocCont, aes(x="",y=Frequency,fill=Clade,)) + 
  geom_bar(stat="identity",width=1) + 
  coord_polar("y", start=0) + 
  theme_void() + 
  scale_fill_brewer(palette="Paired") + 
  facet_wrap(mainPropTableLocCont$Continent) 
#originalCont
```
#By Country
##Prop Table
```{r}
#Original Data Set Proportion Table by Country

mainCladeTableLoc<- table(tb$GISAID_clade, tb$country)
mainPropTableLoc <- as.data.frame(prop.table(mainCladeTableLoc,2)) %>% 
  mutate("Clade"=Var1,
       "Country"=Var2,
         "Frequency"=Freq*100) %>%
  select(Clade,Country,Frequency)
#mainPropTableLoc
```

##Frequency Plot
```{r}
#Original Data Set Percentage by Country

originalLoc <- ggplot(mainPropTableLoc, aes(x="",y=Frequency,fill=Clade,)) + 
  geom_bar(stat="identity",width=1) + 
  coord_polar("y", start=0) + 
  theme_void() + 
  scale_fill_brewer(palette="Paired") + 
  facet_wrap(mainPropTableLoc$Country)
#originalLoc
```
#Undersample 40
#By Month
##Prop Table
```{r}
#Undersample 40 Proportion Table by Month

under40DateTableMonth <- table(undersample40$GISAID_clade, undersample40$month)
under40PropTableDateMonth <- as.data.frame(prop.table(under40DateTableMonth,2)) %>% 
  mutate("Clade"=Var1,
         "Month"=Var2,
         "Frequency"=Freq*100) %>%
  select(Clade,Month,Frequency)
#under40PropTableDateMonth
```

##Frequency Plot
```{r}
#Undersample 40 Frequency Percentage Plot Month

under40DateMonth <- ggplot(under40PropTableDateMonth,aes(x=Month,y=Frequency,fill=Clade, width=6)) + 
  geom_col() + 
  scale_x_discrete(limits=under40PropTableDateMonth$Month) + 
  scale_fill_brewer(palette="Paired") + 
  theme_bw()
#under40DateMonth
```

#By Week
##Prop Table
```{r}
#Undersample 40 Proportion Table by Week

under40DateTableWeek <- table(undersample40$GISAID_clade, undersample40$week)
under40PropTableDateWeek <- as.data.frame(prop.table(under40DateTableWeek,2)) %>% 
  mutate("Clade"=Var1,
         "Week"=Var2,
         "Frequency"=Freq*100) %>%
  select(Clade,Week,Frequency)
#under40PropTableDateWeek
```

##Frequency Plot
```{r}
#Undersample 40 Frequency Percentage Plot
under40DateWeek <- ggplot(under40PropTableDateWeek,aes(x=Week,y=Frequency,fill=Clade,width=6)) + 
  geom_col() + 
  scale_x_discrete(limits=under40PropTableDateWeek$Week) + 
  scale_fill_brewer(palette="Paired") + 
  theme_bw()
#under40DateWeek
```
#By Continent
##Prop Table
```{r}
#Undersample 40 Proportion Table by Continent
under40CladeTableLocCont <- table(undersample40$GISAID_clade, undersample40$region)
under40PropTableLocCont <- as.data.frame(prop.table(under40CladeTableLocCont,2)) %>% 
  mutate("Clade"=Var1,
       "Continent"=Var2,
         "Frequency"=Freq*100) %>%
  select(Clade,Continent,Frequency)
#under40PropTableLocCont
```

##Frequency Plot
```{r}
#Undersample 40 Percentage by Continent

under40Cont <- ggplot(under40PropTableLocCont, aes(x="",y=Frequency,fill=Clade,)) + 
  geom_bar(stat="identity",width=1) + 
  coord_polar("y", start=0) + 
  theme_void() + 
  scale_fill_brewer(palette="Paired") + 
  facet_wrap(under40PropTableLocCont$Continent) 
#under40Cont
```

#By Country
##Prop Table
```{r}
#Undersample 40 Proportion Table by Country

under40CladeTableLoc <- table(undersample40$GISAID_clade, undersample40$country)
under40PropTableLoc <- as.data.frame(prop.table(under40CladeTableLoc,2)) %>% 
  mutate("Clade"=Var1,
       "Country"=Var2,
         "Frequency"=Freq*100) %>%
  select(Clade,Country,Frequency)
#under40PropTableLoc
```

##Frequency Plot
```{r}
#Undersample 40 Percentage by Country

under40Loc <- ggplot(under40PropTableLoc, aes(x="",y=Frequency,fill=Clade)) + 
  geom_bar(stat="identity",width=1) + 
  coord_polar("y", start=0) + 
  theme_void() +
  scale_fill_brewer(palette="Paired") + 
  facet_wrap(under40PropTableLoc$Country)
#under40Loc
```

#Undersample of 20
#By Month
##Prop Table
```{r}
#Undersample 20 Proportion Table by Month

under20DateTableMonth <- table(undersample20$GISAID_clade, undersample20$month)
under20PropTableDateMonth <- as.data.frame(prop.table(under20DateTableMonth,2)) %>% 
  mutate("Clade"=Var1,
         "Month"=Var2,
         "Frequency"=Freq*100) %>%
  select(Clade,Month,Frequency)
#under20PropTableDateMonth
```

##Frequency Plot
```{r}
#Undersample 20 Frequency Percentage Plot Month

under20DateMonth <- ggplot(under20PropTableDateMonth,aes(x=Month,y=Frequency,fill=Clade, width=6)) + 
  geom_col() + 
  scale_x_discrete(limits=under20PropTableDateMonth$Month) + 
  scale_fill_brewer(palette="Paired") + 
  theme_bw()
#under20DateMonth
```

#By Week
##Prop Table
```{r}
#Undersample 20 Proportion Table by Week

under20DateTableWeek <- table(undersample20$GISAID_clade, undersample20$week)
under20PropTableDateWeek <- as.data.frame(prop.table(under20DateTableWeek,2)) %>% 
  mutate("Clade"=Var1,
         "Week"=Var2,
         "Frequency"=Freq*100) %>%
  select(Clade,Week,Frequency)
#under20PropTableDateWeek
```

##Frequency Plot
```{r}
#Undersample 20 Frequency Percentage Plot

under20DateWeek <- ggplot(under20PropTableDateWeek,aes(x=Week,y=Frequency,fill=Clade,width=6)) + 
  geom_col() + 
  scale_x_discrete(limits=under20PropTableDateWeek$Week) + 
  scale_fill_brewer(palette="Paired") + 
  theme_bw()
#under20DateWeek
```
#By Continent
##Prop Table
```{r}
#Undersample 20 Proportion Table by Continent

under20CladeTableLocCont <- table(undersample20$GISAID_clade, undersample20$region)
under20PropTableLocCont <- as.data.frame(prop.table(under20CladeTableLocCont,2)) %>% 
  mutate("Clade"=Var1,
       "Continent"=Var2,
         "Frequency"=Freq*100) %>%
  select(Clade,Continent,Frequency)
#under20PropTableLocCont
```
##Frequency Plot
```{r}
#Undersample 20 Percentage by Continent

under20Cont <- ggplot(under20PropTableLocCont, aes(x="",y=Frequency,fill=Clade,)) + 
  geom_bar(stat="identity",width=1) + 
  coord_polar("y", start=0) + 
  theme_void() + 
  scale_fill_brewer(palette="Paired") + 
  facet_wrap(under20PropTableLocCont$Continent) 
#under20Cont
```

#By Country
##Prop Table
```{r}
#Undersample 20 Proportion Table by Country

under20CladeTableLoc <- table(undersample20$GISAID_clade, undersample20$country)
under20PropTableLoc <- as.data.frame(prop.table(under20CladeTableLoc,2)) %>% 
  mutate("Clade"=Var1,
       "Country"=Var2,
         "Frequency"=Freq*100) %>%
  select(Clade,Country,Frequency)
#under20PropTableLoc
```
##Frequency Plot
```{r}
#Undersample 20 Percentage by Country

under20Loc <- ggplot(under20PropTableLoc, aes(x="",y=Frequency,fill=Clade)) + 
  geom_bar(stat="identity",width=1) + 
  coord_polar("y", start=0) + 
  theme_void() +
  scale_fill_brewer(palette="Paired") + 
  facet_wrap(under20PropTableLoc$Country)
#under20Loc
```

#Oversample of 200
#By Month
##Prop Table
```{r}
#Oversample 200 Proportion Table by Month

over200DateTableMonth <- table(oversample200$GISAID_clade, oversample200$month)
over200PropTableDateMonth <- as.data.frame(prop.table(over200DateTableMonth,2)) %>% 
  mutate("Clade"=Var1,
         "Month"=Var2,
         "Frequency"=Freq*100) %>%
  select(Clade,Month,Frequency)
#over200PropTableDateMonth
```

##Frequency Plot
```{r}
#Oversample 200 Frequency Percentage Plot Month

over200DateMonth <- ggplot(over200PropTableDateMonth,aes(x=Month,y=Frequency,fill=Clade, width=6)) + 
  geom_col() + 
  scale_x_discrete(limits=over200PropTableDateMonth$Month) + 
  scale_fill_brewer(palette="Paired") + 
  theme_bw()
#over200DateMonth
```
#By Week
##Prop Table
```{r}
#Oversample 200 Proportion Table by Week

over200DateTableWeek <- table(oversample200$GISAID_clade, oversample200$week)
over200PropTableDateWeek <- as.data.frame(prop.table(over200DateTableWeek,2)) %>% 
  mutate("Clade"=Var1,
         "Week"=Var2,
         "Frequency"=Freq*100) %>%
  select(Clade,Week,Frequency)
#over200PropTableDateWeek
```

##Frequency Plot
```{r}
#Oversample 200 Frequency Percentage Plot Week

over200DateWeek <- ggplot(over200PropTableDateWeek,aes(x=Week,y=Frequency,fill=Clade,width=6)) + 
  geom_col() + 
  scale_x_discrete(limits=over200PropTableDateWeek$Week) + 
  scale_fill_brewer(palette="Paired") + 
  theme_bw()
#over200DateWeek
```
#By Continent
##Prop Table
```{r}
#Oversample 200 Proportion Table by Continent

over200CladeTableLocCont <- table(oversample200$GISAID_clade, oversample200$region)
over200PropTableLocCont <- as.data.frame(prop.table(over200CladeTableLocCont,2)) %>% 
  mutate("Clade"=Var1,
       "Continent"=Var2,
         "Frequency"=Freq*100) %>%
  select(Clade,Continent,Frequency)
#over200PropTableLocCont
```
##Frequency Plot
```{r}
#Oversample 200 Percentage by Continent

over200Cont <- ggplot(over200PropTableLocCont, aes(x="",y=Frequency,fill=Clade,)) + 
  geom_bar(stat="identity",width=1) + 
  coord_polar("y", start=0) + 
  theme_void() + 
  scale_fill_brewer(palette="Paired") + 
  facet_wrap(over200PropTableLocCont$Continent) 
#over200Cont
```

#By Country
##Prop Table
```{r}
#Oversample 200 Proportion Table by Country

over200CladeTableLoc <- table(oversample200$GISAID_clade, oversample200$country)
over200PropTableLoc <- as.data.frame(prop.table(over200CladeTableLoc,2)) %>% 
  mutate("Clade"=Var1,
       "Country"=Var2,
         "Frequency"=Freq*100) %>%
  select(Clade,Country,Frequency)
#over200PropTableLoc
```
##Frequency Plot
```{r}
#Oversample 200 Percentage by Country

over200Loc <- ggplot(over200PropTableLoc, aes(x="",y=Frequency,fill=Clade)) + 
  geom_bar(stat="identity",width=1) + 
  coord_polar("y", start=0) + 
  theme_void() + 
  scale_fill_brewer(palette="Paired") + 
  facet_wrap(over200PropTableLoc$Country)
#over200Loc
```

#Oversample of 500
#By Month
##Prop Table
```{r}
#Oversample 500 Proportion Table by Month

over500DateTableMonth <- table(oversample500$GISAID_clade, oversample500$month)
over500PropTableDateMonth <- as.data.frame(prop.table(over500DateTableMonth,2)) %>% 
  mutate("Clade"=Var1,
       "Month"=Var2,
         "Frequency"=Freq*100) %>%
  select(Clade,Month,Frequency)
#over500PropTableDateMonth
```

##Frequency Plot
```{r}
#Oversample 500 Frequency Percentage Plot Month

over500DateMonth <- ggplot(over500PropTableDateMonth,aes(x=Month,y=Frequency,fill=Clade, width=6)) + 
  geom_col() + 
  scale_x_discrete(limits=over500PropTableDateMonth$Month) + 
  scale_fill_brewer(palette="Paired") + 
  theme_bw()
#over500DateMonth
```

#By Week
##Prop Table
```{r}
#Oversample 500 Proportion Table by Week

over500DateTableWeek <- table(oversample500$GISAID_clade, oversample500$week)
over500PropTableDateWeek <- as.data.frame(prop.table(over500DateTableWeek,2)) %>% 
  mutate("Clade"=Var1,
         "Week"=Var2,
         "Frequency"=Freq*100) %>%
  select(Clade,Week,Frequency)
#over500PropTableDateWeek
```

##Frequency Plot
```{r}
#Oversample 500 Frequency Percentage Plot Week

over500DateWeek <- ggplot(over500PropTableDateWeek,aes(x=Week,y=Frequency,fill=Clade,width=6)) + 
  geom_col() + 
  scale_x_discrete(limits=over500PropTableDateWeek$Week) + 
  scale_fill_brewer(palette="Paired") + 
  theme_bw()
#over500DateWeek
```

#By Continent
##Prop Table
```{r}
#Oversample 500 Proportion Table by Country

over500CladeTableLocCont <- table(oversample500$GISAID_clade, oversample500$region)
over500PropTableLocCont <- as.data.frame(prop.table(over500CladeTableLocCont,2)) %>% 
  mutate("Clade"=Var1,
       "Continent"=Var2,
         "Frequency"=Freq*100) %>%
  select(Clade,Continent,Frequency)
#over500PropTableLocCont
```

##Frequency Plot
```{r}
#Oversample 500 Percentage by Month

over500Cont <- ggplot(over500PropTableLocCont, aes(x="",y=Frequency,fill=Clade,)) + 
  geom_bar(stat="identity",width=1) + 
  coord_polar("y", start=0) + 
  theme_void() + 
  scale_fill_brewer(palette="Paired") + 
  facet_wrap(over500PropTableLocCont$Continent) 
#over500Cont
```

#By Country
##Prop Table
```{r}
#Oversample 500 Proportion Table by Country

over500CladeTableLoc <- table(oversample500$GISAID_clade, oversample500$country)
over500PropTableLoc <- as.data.frame(prop.table(over500CladeTableLoc,2)) %>% 
  mutate("Clade"=Var1,
       "Country"=Var2,
         "Frequency"=Freq*100) %>%
  select(Clade,Country,Frequency)
#over500PropTableLoc
```

##Frequency Plot
```{r}
#Oversample 500 Percentage by Month

over500Loc <- ggplot(over500PropTableLoc, aes(x="",y=Frequency,fill=Clade)) + 
  geom_bar(stat="identity",width=1) + 
  coord_polar("y", start=0) + 
  theme_void() + 
  scale_fill_brewer(palette="Paired") + 
  facet_wrap(over500PropTableLoc$Country)
#over500Loc
```

# Date Frequency Plots

## Column {.tabset}

### Original Data
```{r}
#Print all Month Frequency Plots

originalDateMonth
```

### Undersample of 20
```{r}
under20DateMonth
```

### Undersample of 40
```{r}
under40DateMonth
```

### Oversample of 200
```{r}
over200DateMonth
```

### Oversample of 500
```{r}
over500DateMonth
```

## Column {.tabset}

### Original Data
```{r}
#Print all Week Frequency Plots

originalDateWeek
```

### Undersample of 20
```{r}
under20DateWeek
```

### Undersample of 40
```{r}
under40DateWeek 
```

### Oversample of 200
```{r}
over200DateWeek
```

### Oversample 500
```{r}
over500DateWeek
```

# Frequency Plots by Location

## Column {.tabset}

### Original Data
```{r}
#Print all Continent Frequency Plots

originalCont
```

### Undersample of 20
```{r}
under20Cont
```

### Undersample of 40
```{r}
under40Cont
```

### Oversample 200
```{r}
over200Cont
```

### Oversample 500
```{r}
over500Cont
```

## Column {.tabset}

### Original Data
```{r}
#Print all Country Frequency Plots

originalLoc
```

### Undersample of 20
```{r}
under20Loc
```

### Undersample 40
```{r}
under40Loc
```

### Oversample of 200
```{r}
over200Loc
```

### Oversample of 500
```{r}
over500Loc
```

# Chi Squared Tests of Undersamples

## Column {.tabset}

### Month
```{r warning=FALSE}
#Creating sub tables undersample 20 Month

originalCladeTableMonth <- as.data.frame(table(tb$GISAID_clade,tb$month)) %>% mutate("Clade"=Var1,"Month"=Var2) %>% 
  select(Clade,Month,Freq)
under20CladeTableMonth <- as.data.frame(table(undersample20$GISAID_clade,undersample20$month))
compareMonths <- cbind2(originalCladeTableMonth,mainPropTableMonth$Frequency) %>% 
  mutate("Prop"=y/100,"Original"=Freq) %>% select(Clade,Month,Original,Prop) %>% 
  filter(Month %in% under20CladeTableMonth$Var2)

#Combining tables and renaming variables

compareMonths <- cbind2(compareMonths,under20CladeTableMonth$Freq) %>% group_by(Month) %>% 
  mutate("N_under20"=sum(y), 
         "Observed_under20"=y,
         "Expected_under20"=round(N_under20*Prop)) %>% 
  select(-y)
#compareMonths

#Chi-squared tests for undersample 20 Months

list20_1 <- split(compareMonths,compareMonths$Month, drop=TRUE)
month20pVals <- as.data.frame(matrix(nrow=length(list20_1),ncol=3))
colnames(month20pVals) <- c("Month","p-Value","DF")
for (i in 1:length(list20_1)){
  temp <- as.data.frame(list20_1[i])
  colnames(temp) <- c("Clade","Month","Original","Prop","N_under20","Observed_under20","Expected_under20")
  temp <- temp %>% filter(Expected_under20!=0)
  if (nrow(temp)>=4){
    originalChiMonth20 <- chisq.test(temp$Observed_under20,correct=TRUE,p=temp$Expected_under20,rescale.p=TRUE)
    #Store chi squared tests in data frame
    #print(originalChiMonth20)
    temp <- data.frame(lapply(temp, as.character), stringsAsFactors=FALSE)
    month20pVals[i,1] <- temp[1,2]
    month20pVals[i,2] <- originalChiMonth20$p.value
    month20pVals[i,3] <- originalChiMonth20$parameter
  }
}
month20pVals <- month20pVals %>% drop_na
month20pVals
```

### Week
```{r warning=FALSE}
#Creating sub tables undersample 20 Week

originalCladeTableWeek <- as.data.frame(table(tb$GISAID_clade,tb$week)) %>% mutate("Clade"=Var1,"Week"=Var2) %>% 
  select(Clade,Week,Freq)
under20CladeTableWeek <- as.data.frame(table(undersample20$GISAID_clade,undersample20$week))
compareWeeks <- cbind2(originalCladeTableWeek,mainPropTableWeek$Frequency) %>%
  mutate("Prop"=y/100,"Original"=Freq) %>% select(Clade,Week,Original,Prop) %>% 
  filter(Week %in% under20CladeTableWeek$Var2)

#Combining tables and renaming variables

compareWeeks <- cbind2(compareWeeks,under20CladeTableWeek$Freq) %>% group_by(Week) %>% 
  mutate("N_under20"=sum(y), 
         "Observed_under20"=y,
         "Expected_under20"=round(N_under20*Prop)) %>% 
  select(-y)
#compareWeeks

#Chi-squared tests for undersample 20 Week

list20_2 <- split(compareWeeks,compareWeeks$Week, drop=TRUE)
week20pVals <- as.data.frame(matrix(nrow=length(list20_2),ncol=3))
colnames(week20pVals) <- c("Week","p-Value","DF")
for (i in 1:length(list20_2)){
  temp <- as.data.frame(list20_2[i])
  colnames(temp) <- c("Clade","Week","Original","Prop","N_under20","Observed_under20","Expected_under20")
  temp <- temp %>% filter(Expected_under20!=0)
  if (nrow(temp)>=4){
    originalChiWeek20 <- chisq.test(temp$Observed_under20,correct=TRUE,p=temp$Expected_under20,rescale.p=TRUE)
    #Store chi squared tests in data frame
    #print(originalChiWeek20)
    temp <- data.frame(lapply(temp, as.character), stringsAsFactors=FALSE)
    week20pVals[i,1] <- temp[1,2]
    week20pVals[i,2] <- originalChiWeek20$p.value
    week20pVals[i,3] <- originalChiWeek20$parameter
  }
}
week20pVals <- week20pVals %>% drop_na
week20pVals
```

### Continent
```{r warning=FALSE}
#Creating sub tables undersample 20 Continent

originalCladeTableCont <- as.data.frame(table(tb$GISAID_clade,tb$region)) %>% mutate("Clade"=Var1,"Continent"=Var2) %>% 
  select(Clade,Continent,Freq)
under20CladeTableCont <- as.data.frame(table(undersample20$GISAID_clade,undersample20$region))
compareCont <- cbind2(originalCladeTableCont,mainPropTableLocCont$Frequency) %>% 
  mutate("Prop"=y/100,"Original"=Freq) %>% select(Clade,Continent,Original,Prop) %>% 
  filter(Continent %in% under20CladeTableCont$Var2)

#Combining tables and renaming variables

compareCont <- cbind2(compareCont,under20CladeTableCont$Freq) %>% group_by(Continent) %>% 
  mutate("N_under20"=sum(y), 
         "Observed_under20"=y,
         "Expected_under20"=round(N_under20*Prop)) %>% 
  select(-y)
#compareCont

#Chi-squared tests for undersample 20 Continents

list20_3 <- split(compareCont,compareCont$Continent, drop=TRUE)
cont20pVals <- as.data.frame(matrix(nrow=length(list20_3),ncol=3))
colnames(cont20pVals) <- c("Continent","p-Value","DF")
for (i in 1:length(list20_3)){
  temp <- as.data.frame(list20_3[i])
  colnames(temp) <- c("Clade","Continent","Original","Prop","N_under20","Observed_under20","Expected_under20")
  temp <- temp %>% filter(Expected_under20!=0)
  if (nrow(temp)>=4){
    originalChiCont20 <- chisq.test(temp$Observed_under20,correct=TRUE,p=temp$Expected_under20,rescale.p=TRUE)
    #Store chi squared tests in data frame
    #print(originalChiCont20)
    temp <- data.frame(lapply(temp, as.character), stringsAsFactors=FALSE)
    cont20pVals[i,1] <- temp[1,2]
    cont20pVals[i,2] <- originalChiCont20$p.value
    cont20pVals[i,3] <- originalChiCont20$parameter
  }
}
cont20pVals <- cont20pVals %>% drop_na
cont20pVals
```

### Country
```{r warning=FALSE}
#Creating sub tables undersample 20 Country

originalCladeTableLoc <- as.data.frame(table(tb$GISAID_clade,tb$country)) %>% mutate("Clade"=Var1,"Country"=Var2) %>% 
  select(Clade,Country,Freq)
under20CladeTableLoc <- as.data.frame(table(undersample20$GISAID_clade,undersample20$country))
compareLoc <- cbind2(originalCladeTableLoc,mainPropTableLoc$Frequency) %>%      
  mutate("Prop"=y/100,"Original"=Freq) %>% select(Clade,Country,Original,Prop) %>% 
  filter(Country %in% under20CladeTableLoc$Var2)

#Combining tables and renaming variables

compareLoc <- cbind2(compareLoc,under20CladeTableLoc$Freq) %>% group_by(Country) %>% 
  mutate("N_under20"=sum(y), 
         "Observed_under20"=y,
         "Expected_under20"=round(N_under20*Prop)) %>% 
  select(-y)
#compareLoc

#Chi-squared tests for undersample 20 Countries

list20_4 <- split(compareLoc,compareLoc$Country, drop=TRUE)
country20pVals <- as.data.frame(matrix(nrow=length(list20_4),ncol=3))
colnames(country20pVals) <- c("Country","p-Value","DF")
for (i in 1:length(list20_4)){
  temp <- as.data.frame(list20_4[i])
  colnames(temp) <- c("Clade","Country","Original","Prop","N_under20","Observed_under20","Expected_under20")
  temp <- temp %>% filter(Expected_under20!=0)
  if (nrow(temp)>=4){
    originalChiLoc20 <- chisq.test(temp$Observed_under20,correct=TRUE,p=temp$Expected_under20,rescale.p=TRUE)
    #Store chi squared tests in data frame
    #print(originalChiLoc20)
    temp <- data.frame(lapply(temp, as.character), stringsAsFactors=FALSE)
    country20pVals[i,1] <- temp[1,2]
    country20pVals[i,2] <- originalChiLoc20$p.value
    country20pVals[i,3] <- originalChiLoc20$parameter
  }
}
country20pVals <- country20pVals %>% drop_na
country20pVals
```

## Column {.tabset}

### Month
```{r warning=FALSE}
#Creating sub tables undersample 40 Month

originalCladeTableMonth <- as.data.frame(table(tb$GISAID_clade,tb$month)) %>% mutate("Clade"=Var1,"Month"=Var2) %>% 
  select(Clade,Month,Freq)
under40CladeTableMonth <- as.data.frame(table(undersample40$GISAID_clade,undersample40$month))
#under40CladeTableMonth
compare40Months <- cbind2(originalCladeTableMonth,mainPropTableMonth$Frequency) %>% 
  mutate("Prop"=y/100,"Original"=Freq) %>% select(Clade,Month,Original,Prop) %>% 
  filter(Month %in% under40CladeTableMonth$Var2)

#Combining tables and renaming variables

compare40Months <- cbind2(compare40Months,under40CladeTableMonth$Freq) %>% group_by(Month) %>% 
  mutate("N_under40"=sum(y), 
         "Observed_under40"=y,
         "Expected_under40"=round(N_under40*Prop)) %>% 
  select(-y)
#compare40Months

#Chi-squared tests for undersample 40 Months

list40_1 <- split(compare40Months,compare40Months$Month, drop=TRUE)
month40pVals <- as.data.frame(matrix(nrow=length(list40_1),ncol=3))
colnames(month40pVals) <- c("Month","p-Value","DF")
for (i in 1:length(list40_1)){
  temp <- as.data.frame(list40_1[i])
  colnames(temp) <- c("Clade","Month","Original","Prop","N_under40","Observed_under40","Expected_under40")
  temp <- temp %>% filter(Expected_under40!=0)
  if (nrow(temp)>=4){
    originalChiMonth40 <- chisq.test(temp$Observed_under40,correct=TRUE,p=temp$Expected_under40,rescale.p=TRUE)
    #Store chi squared tests in data frame
    #print(originalChiMonth40)
    temp <- data.frame(lapply(temp, as.character), stringsAsFactors=FALSE)
    month40pVals[i,1] <- temp[1,2]
    month40pVals[i,2] <- originalChiMonth40$p.value
    month40pVals[i,3] <- originalChiMonth40$parameter
  }
}
month40pVals <- month40pVals %>% drop_na
month40pVals
```

### Week
```{r warning=FALSE}
#Creating sub tables undersample 40 Week

originalCladeTableWeek <- as.data.frame(table(tb$GISAID_clade,tb$week)) %>% mutate("Clade"=Var1,"Week"=Var2) %>% 
  select(Clade,Week,Freq)
under40CladeTableWeek <- as.data.frame(table(undersample40$GISAID_clade,undersample40$week))
compare40Weeks <- cbind2(originalCladeTableWeek,mainPropTableWeek$Frequency) %>%
  mutate("Prop"=y/100,"Original"=Freq) %>% select(Clade,Week,Original,Prop) %>% 
  filter(Week %in% under40CladeTableWeek$Var2)

#Combining tables and renaming variables

compare40Weeks <- cbind2(compare40Weeks,under40CladeTableWeek$Freq) %>% group_by(Week) %>% 
  mutate("N_under40"=sum(y), 
         "Observed_under40"=y,
         "Expected_under40"=round(N_under40*Prop)) %>% 
  select(-y)
#compare40Weeks

#Chi-squared tests for undersample 40 Weeks

list40_2 <- split(compare40Weeks,compare40Weeks$Week, drop=TRUE)
week40pVals <- as.data.frame(matrix(nrow=length(list40_2),ncol=3))
colnames(week40pVals) <- c("Week","p-Value","DF")
for (i in 1:length(list40_2)){
  temp <- as.data.frame(list40_2[i])
  colnames(temp) <- c("Clade","Week","Original","Prop","N_under40","Observed_under40","Expected_under40")
  temp <- temp %>% filter(Expected_under40!=0)
  if (nrow(temp)>=4){
    originalChiWeek40 <- chisq.test(temp$Observed_under40,correct=TRUE,p=temp$Expected_under40,rescale.p=TRUE)
    #Store chi squared tests in data frame
    #print(originalChiWeek40)
    temp <- data.frame(lapply(temp, as.character), stringsAsFactors=FALSE)
    week40pVals[i,1] <- temp[1,2]
    week40pVals[i,2] <- originalChiWeek40$p.value
    week40pVals[i,3] <- originalChiWeek40$parameter
  }
}
week40pVals <- week40pVals %>% drop_na
week40pVals
```

### Continent
```{r warning=FALSE}
#Creating sub tables undersample 40 Continent

originalCladeTableCont <- as.data.frame(table(tb$GISAID_clade,tb$region)) %>% mutate("Clade"=Var1,"Continent"=Var2) %>% 
  select(Clade,Continent,Freq)
under40CladeTableCont <- as.data.frame(table(undersample40$GISAID_clade,undersample40$region))
compare40Cont <- cbind2(originalCladeTableCont,mainPropTableLocCont$Frequency) %>% 
  mutate("Prop"=y/100,"Original"=Freq) %>% select(Clade,Continent,Original,Prop) %>% 
  filter(Continent %in% under40CladeTableCont$Var2)

#Combining tables and renaming variables

compare40Cont <- cbind2(compare40Cont,under40CladeTableCont$Freq) %>% group_by(Continent) %>% 
  mutate("N_under40"=sum(y), 
         "Observed_under40"=y,
         "Expected_under40"=round(N_under40*Prop)) %>% 
  select(-y)
#compare40Cont

#Chi-squared tests for undersample 40 Continents

list40_3 <- split(compare40Cont,compare40Cont$Continent, drop=TRUE)
cont40pVals <- as.data.frame(matrix(nrow=length(list40_3),ncol=3))
colnames(cont40pVals) <- c("Continent","p-Value","DF")
for (i in 1:length(list40_3)){
  temp <- as.data.frame(list40_3[i])
  colnames(temp) <- c("Clade","Continent","Original","Prop","N_under40","Observed_under40","Expected_under40")
  temp <- temp %>% filter(Expected_under40!=0)
  if (nrow(temp)>=4){
    originalChiCont40 <- chisq.test(temp$Observed_under40,correct=TRUE,p=temp$Expected_under40,rescale.p=TRUE)
    #Store chi squared tests in data frame
    #print(originalChiCont40)
    temp <- data.frame(lapply(temp, as.character), stringsAsFactors=FALSE)
    cont40pVals[i,1] <- temp[1,2]
    cont40pVals[i,2] <- originalChiCont40$p.value
    cont40pVals[i,3] <- originalChiCont40$parameter
  }
}
cont40pVals <- cont40pVals %>% drop_na
cont40pVals
```

### Country
```{r warning=FALSE}
#Creating sub tables undersample 40 Country

originalCladeTableLoc <- as.data.frame(table(tb$GISAID_clade,tb$country)) %>% mutate("Clade"=Var1,"Country"=Var2) %>% 
  select(Clade,Country,Freq)
under40CladeTableLoc <- as.data.frame(table(undersample40$GISAID_clade,undersample40$country))
compare40Loc <- cbind2(originalCladeTableLoc,mainPropTableLoc$Frequency) %>% 
  mutate("Prop"=y/100,"Original"=Freq) %>% select(Clade,Country,Original,Prop) %>% 
  filter(Country %in% under40CladeTableLoc$Var2)

#Combining tables and renaming variables

compare40Loc <- cbind2(compare40Loc,under40CladeTableLoc$Freq) %>% group_by(Country) %>% 
  mutate("N_under40"=sum(y), 
         "Observed_under40"=y,
         "Expected_under40"=round(N_under40*Prop)) %>% 
  select(-y)
#compare40Loc

#Chi-squared tests for undersample 40 Countries

list40_4 <- split(compare40Loc,compare40Loc$Country, drop=TRUE)
country40pVals <- as.data.frame(matrix(nrow=length(list40_4),ncol=3))
colnames(country40pVals) <- c("Country","p-Value","DF")
for (i in 1:length(list40_4)){
  temp <- as.data.frame(list40_4[i])
  colnames(temp) <- c("Clade","Country","Original","Prop","N_under40","Observed_under40","Expected_under40")
  temp <- temp %>% filter(Expected_under40!=0)
  if (nrow(temp)>=4){
    originalChiLoc40 <- chisq.test(temp$Observed_under40,correct=TRUE,p=temp$Expected_under40,rescale.p=TRUE)
    #Store chi squared tests in data frame
    #print(originalChiLoc40)
    temp <- data.frame(lapply(temp, as.character), stringsAsFactors=FALSE)
    country40pVals[i,1] <- temp[1,2]
    country40pVals[i,2] <- originalChiLoc40$p.value
    country40pVals[i,3] <- originalChiLoc40$parameter
  }
}
country40pVals <- country40pVals %>% drop_na
country40pVals
```
# Chi Squared Tests of Oversamples

## Column {.tabset}

### Month
```{r warning=FALSE}
#Creating sub tables oversample 200 Month

originalCladeTableMonth <- as.data.frame(table(tb$GISAID_clade,tb$month)) %>% mutate("Clade"=Var1,"Month"=Var2) %>% 
  select(Clade,Month,Freq)
over200CladeTableMonth <- as.data.frame(table(oversample200$GISAID_clade,oversample200$month))
compare200Months <- cbind2(originalCladeTableMonth,mainPropTableMonth$Frequency) %>% 
  mutate("Prop"=y/100,"Original"=Freq) %>% select(Clade,Month,Original,Prop) %>% 
  filter(Month %in% over200CladeTableMonth$Var2)

#Combining tables and renaming variables

compare200Months <- cbind2(compare200Months,over200CladeTableMonth$Freq) %>% group_by(Month) %>% 
  mutate("N_over200"=sum(y), 
         "Observed_over200"=y,
         "Expected_over200"=round(N_over200*Prop)) %>% 
  select(-y)
#compare200Months

#Chi-squared tests for oversample 200 Months

list200_1 <- split(compare200Months,compare200Months$Month, drop=TRUE)
month200pVals <- as.data.frame(matrix(nrow=length(list200_1),ncol=3))
colnames(month200pVals) <- c("Month","p-Value","DF")
for (i in 1:length(list200_1)){
  temp <- as.data.frame(list200_1[i])
  colnames(temp) <- c("Clade","Month","Original","Prop","N_over200","Observed_over200","Expected_over200")
  temp <- temp %>% filter(Expected_over200!=0)
  if (nrow(temp)>=4){
    originalChiMonth200 <- chisq.test(temp$Observed_over200,correct=TRUE,p=temp$Expected_over200,rescale.p=TRUE)
    #Store chi squared tests in data frame
    #print(originalChiMonth200)
    temp <- data.frame(lapply(temp, as.character), stringsAsFactors=FALSE)
    month200pVals[i,1] <- temp[1,2]
    month200pVals[i,2] <- originalChiMonth200$p.value
    month200pVals[i,3] <- originalChiMonth200$parameter
  }
}
month200pVals <- month200pVals %>% drop_na
month200pVals
```

### Week
```{r warning=FALSE}
#Creating sub tables oversample 200 Week

originalCladeTableWeek <- as.data.frame(table(tb$GISAID_clade,tb$week)) %>% mutate("Clade"=Var1,"Week"=Var2) %>% 
  select(Clade,Week,Freq)
over200CladeTableWeek <- as.data.frame(table(oversample200$GISAID_clade,oversample200$week))
compare200Weeks <- cbind2(originalCladeTableWeek,mainPropTableWeek$Frequency) %>%
  mutate("Prop"=y/100,"Original"=Freq) %>% select(Clade,Week,Original,Prop) %>% 
  filter(Week %in% over200CladeTableWeek$Var2)

#Combining tables and renaming variables

compare200Weeks <- cbind2(compare200Weeks,over200CladeTableWeek$Freq) %>% group_by(Week) %>% 
  mutate("N_over200"=sum(y), 
         "Observed_over200"=y,
         "Expected_over200"=round(N_over200*Prop)) %>% 
  select(-y)
#compare200Weeks

#Chi-squared tests for oversample 200 Weeks

list200_2 <- split(compare200Weeks,compare200Weeks$Week, drop=TRUE)
week200pVals <- as.data.frame(matrix(nrow=length(list200_2),ncol=3))
colnames(week200pVals) <- c("Week","p-Value","DF")
for (i in 1:length(list200_2)){
  temp <- as.data.frame(list200_2[i])
  colnames(temp) <- c("Clade","Week","Original","Prop","N_over200","Observed_over200","Expected_over200")
  temp <- temp %>% filter(Expected_over200!=0)
  if (nrow(temp)>=4){
    originalChiWeek200 <- chisq.test(temp$Observed_over200,correct=TRUE,p=temp$Expected_over200,rescale.p=TRUE)
    #Store chi squared tests in data frame
    #print(originalChiWeek200)
    temp <- data.frame(lapply(temp, as.character), stringsAsFactors=FALSE)
    week200pVals[i,1] <- temp[1,2]
    week200pVals[i,2] <- originalChiWeek200$p.value
    week200pVals[i,3] <- originalChiWeek200$parameter
  }
}
week200pVals <- week200pVals %>% drop_na
week200pVals
```

### Continent
```{r warning=FALSE}
#Creating sub tables oversample 200 Continent

originalCladeTableCont <- as.data.frame(table(tb$GISAID_clade,tb$region)) %>% mutate("Clade"=Var1,"Continent"=Var2) %>% 
  select(Clade,Continent,Freq)
over200CladeTableCont <- as.data.frame(table(oversample200$GISAID_clade,oversample200$region))
compare200Cont <- cbind2(originalCladeTableCont,mainPropTableLocCont$Frequency) %>% 
  mutate("Prop"=y/100,"Original"=Freq) %>% select(Clade,Continent,Original,Prop) %>% 
  filter(Continent %in% over200CladeTableCont$Var2)

#Combining tables and renaming variables

compare200Cont <- cbind2(compare200Cont,over200CladeTableCont$Freq) %>% group_by(Continent) %>% 
  mutate("N_over200"=sum(y), 
         "Observed_over200"=y,
         "Expected_over200"=round(N_over200*Prop)) %>% 
  select(-y)
#compare200Cont

#Chi-squared tests for oversample 200 Continents

list200_3 <- split(compare200Cont,compare200Cont$Continent, drop=TRUE)
cont200pVals <- as.data.frame(matrix(nrow=length(list200_3),ncol=3))
colnames(cont200pVals) <- c("Continent","p-Value","DF")
for (i in 1:length(list200_3)){
  temp <- as.data.frame(list200_3[i])
  colnames(temp) <- c("Clade","Continent","Original","Prop","N_over200","Observed_over200","Expected_over200")
  temp <- temp %>% filter(Expected_over200!=0)
  if (nrow(temp)>=4){
    originalChiCont200 <- chisq.test(temp$Observed_over200,correct=TRUE,p=temp$Expected_over200,rescale.p=TRUE)
    #Store chi squared tests in data frame
    #print(originalChiCont200)
    temp <- data.frame(lapply(temp, as.character), stringsAsFactors=FALSE)
    cont200pVals[i,1] <- temp[1,2]
    cont200pVals[i,2] <- originalChiCont200$p.value
    cont200pVals[i,3] <- originalChiCont200$parameter
  }
}
cont200pVals <- cont200pVals %>% drop_na
cont200pVals
```

### Country
```{r warning=FALSE}
#Creating sub tables oversample 200 Country

originalCladeTableLoc <- as.data.frame(table(tb$GISAID_clade,tb$country)) %>% mutate("Clade"=Var1,"Country"=Var2) %>% 
  select(Clade,Country,Freq)
over200CladeTableLoc <- as.data.frame(table(oversample200$GISAID_clade,oversample200$country))
compare200Loc <- cbind2(originalCladeTableLoc,mainPropTableLoc$Frequency) %>% 
  mutate("Prop"=y/100,"Original"=Freq) %>% select(Clade,Country,Original,Prop) %>% 
  filter(Country %in% over200CladeTableLoc$Var2)

#Combining tables and renaming variables

compare200Loc <- cbind2(compare200Loc,over200CladeTableLoc$Freq) %>% group_by(Country) %>% 
  mutate("N_over200"=sum(y), 
         "Observed_over200"=y,
         "Expected_over200"=round(N_over200*Prop)) %>% 
  select(-y)
#compare200Loc

#Chi-squared tests for oversample 200 Countries

list200_4 <- split(compare200Loc,compare200Loc$Country, drop=TRUE)
country200pVals <- as.data.frame(matrix(nrow=length(list200_4),ncol=3))
colnames(country200pVals) <- c("Country","p-Value","DF")
for (i in 1:length(list200_4)){
  temp <- as.data.frame(list200_4[i])
  colnames(temp) <- c("Clade","Country","Original","Prop","N_over200","Observed_over200","Expected_over200")
  temp <- temp %>% filter(Expected_over200!=0)
  if (nrow(temp)>=4){
    originalChiLoc200 <- chisq.test(temp$Observed_over200,correct=TRUE,p=temp$Expected_over200,rescale.p=TRUE)
    #Store chi squared tests in data frame
    #print(originalChiLoc200)
    temp <- data.frame(lapply(temp, as.character), stringsAsFactors=FALSE)
    country200pVals[i,1] <- temp[1,2]
    country200pVals[i,2] <- originalChiLoc200$p.value
    country200pVals[i,3] <- originalChiLoc200$parameter
  }
}
country200pVals <- country200pVals %>% drop_na
country200pVals
```

## Column {.tabset}

### Month
```{r warning=FALSE}
#Creating sub tables oversample 500 Month

originalCladeTableMonth <- as.data.frame(table(tb$GISAID_clade,tb$month)) %>% mutate("Clade"=Var1,"Month"=Var2) %>% 
  select(Clade,Month,Freq)
over500CladeTableMonth <- as.data.frame(table(oversample500$GISAID_clade,oversample500$month))
compare500Months <- cbind2(originalCladeTableMonth,mainPropTableMonth$Frequency) %>% 
  mutate("Prop"=y/100,"Original"=Freq) %>% select(Clade,Month,Original,Prop) %>% 
  filter(Month %in% over500CladeTableMonth$Var2)

#Combining tables and renaming variables

compare500Months <- cbind2(compare500Months,over500CladeTableMonth$Freq) %>% group_by(Month) %>% 
  mutate("N_over500"=sum(y), 
         "Observed_over500"=y,
         "Expected_over500"=round(N_over500*Prop)) %>% 
  select(-y)
#compare500Months

#Chi-squared tests for oversample 500 Months

list500_1 <- split(compare500Months,compare500Months$Month, drop=TRUE)
month500pVals <- as.data.frame(matrix(nrow=length(list500_1),ncol=3))
colnames(month500pVals) <- c("Month","p-Value","DF")
for (i in 1:length(list500_1)){
  temp <- as.data.frame(list500_1[i])
  colnames(temp) <- c("Clade","Month","Original","Prop","N_over500","Observed_over500","Expected_over500")
  temp <- temp %>% filter(Expected_over500!=0)
  if (nrow(temp)>=4){
    originalChiMonth500 <- chisq.test(temp$Observed_over500,correct=TRUE,p=temp$Expected_over500,rescale.p=TRUE)
    #Store chi squared tests in data frame
    #print(originalChiMonth500)
    temp <- data.frame(lapply(temp, as.character), stringsAsFactors=FALSE)
    month500pVals[i,1] <- temp[1,2]
    month500pVals[i,2] <- originalChiMonth500$p.value
    month500pVals[i,3] <- originalChiMonth500$parameter
  }
}
month500pVals <- month500pVals %>% drop_na
month500pVals
```

### Week
```{r warning=FALSE}
#Creating sub tables oversample 500 Week

originalCladeTableWeek <- as.data.frame(table(tb$GISAID_clade,tb$week)) %>% mutate("Clade"=Var1,"Week"=Var2) %>% 
  select(Clade,Week,Freq)
over500CladeTableWeek <- as.data.frame(table(oversample500$GISAID_clade,oversample500$week))
compare500Weeks <- cbind2(originalCladeTableWeek,mainPropTableWeek$Frequency) %>%
  mutate("Prop"=y/100,"Original"=Freq) %>% select(Clade,Week,Original,Prop) %>% 
  filter(Week %in% over500CladeTableWeek$Var2)

#Combining tables and renaming variables

compare500Weeks <- cbind2(compare500Weeks,over500CladeTableWeek$Freq) %>% group_by(Week) %>% 
  mutate("N_over500"=sum(y), 
         "Observed_over500"=y,
         "Expected_over500"=round(N_over500*Prop)) %>% 
  select(-y)
#compare500Weeks

#Chi-squared tests for oversample 500 Weeks

list500_2 <- split(compare500Weeks,compare500Weeks$Week, drop=TRUE)
week500pVals <- as.data.frame(matrix(nrow=length(list500_2),ncol=3))
colnames(week500pVals) <- c("Week","p-Value","DF")
for (i in 1:length(list500_2)){
  temp <- as.data.frame(list500_2[i])
  colnames(temp) <- c("Clade","Week","Original","Prop","N_over500","Observed_over500","Expected_over500")
  temp <- temp %>% filter(Expected_over500!=0)
  if (nrow(temp)>=4){
    originalChiWeek500 <- chisq.test(temp$Observed_over500,correct=TRUE,p=temp$Expected_over500,rescale.p=TRUE)
    #Store chi squared tests in data frame
    #print(originalChiWeek500)
    temp <- data.frame(lapply(temp, as.character), stringsAsFactors=FALSE)
    week500pVals[i,1] <- temp[1,2]
    week500pVals[i,2] <- originalChiWeek500$p.value
    week500pVals[i,3] <- originalChiWeek500$parameter
  }
}
week500pVals <- week500pVals %>% drop_na
week500pVals
```

### Continent
```{r warning=FALSE}
#Creating sub tables oversample 500 Continent

originalCladeTableCont <- as.data.frame(table(tb$GISAID_clade,tb$region)) %>% mutate("Clade"=Var1,"Continent"=Var2) %>% 
  select(Clade,Continent,Freq)
over500CladeTableCont <- as.data.frame(table(oversample500$GISAID_clade,oversample500$region))
compare500Cont <- cbind2(originalCladeTableCont,mainPropTableLocCont$Frequency) %>%
  mutate("Prop"=y/100,"Original"=Freq) %>% select(Clade,Continent,Original,Prop) %>% 
  filter(Continent %in% over500CladeTableCont$Var2)

#Combining tables and renaming variables

compare500Cont <- cbind2(compare500Cont,over500CladeTableCont$Freq) %>% group_by(Continent) %>% 
  mutate("N_over500"=sum(y), 
         "Observed_over500"=y,
         "Expected_over500"=round(N_over500*Prop)) %>% 
  select(-y)
#compare500Cont

#Chi-squared tests for oversample 500 Continents

list500_3 <- split(compare500Cont,compare500Cont$Continent, drop=TRUE)
cont500pVals <- as.data.frame(matrix(nrow=length(list500_3),ncol=3))
colnames(cont500pVals) <- c("Continent","p-Value","DF")
for (i in 1:length(list500_3)){
  temp <- as.data.frame(list500_3[i])
  colnames(temp) <- c("Clade","Continent","Original","Prop","N_over500","Observed_over500","Expected_over500")
  temp <- temp %>% filter(Expected_over500!=0)
  if (nrow(temp)>=4){
    originalChiCont500 <- chisq.test(temp$Observed_over500,correct=TRUE,p=temp$Expected_over500,rescale.p=TRUE)
    #Store chi squared tests in data frame
    #print(originalChiCont500)
    temp <- data.frame(lapply(temp, as.character), stringsAsFactors=FALSE)
    cont500pVals[i,1] <- temp[1,2]
    cont500pVals[i,2] <- originalChiCont500$p.value
    cont500pVals[i,3] <- originalChiCont500$parameter
  }
}
cont500pVals <- cont500pVals %>% drop_na
cont500pVals
```

### Country
```{r warning=FALSE}
#Creating sub tables oversample 500 Country

originalCladeTableLoc <- as.data.frame(table(tb$GISAID_clade,tb$country)) %>% mutate("Clade"=Var1,"Country"=Var2) %>% 
  select(Clade,Country,Freq)
over500CladeTableLoc <- as.data.frame(table(oversample500$GISAID_clade,oversample500$country))
compare500Loc <- cbind2(originalCladeTableLoc,mainPropTableLoc$Frequency) %>% 
  mutate("Prop"=y/100,"Original"=Freq) %>% select(Clade,Country,Original,Prop) %>% 
  filter(Country %in% over500CladeTableLoc$Var2)

#Combining tables and renaming variables

compare500Loc <- cbind2(compare500Loc,over500CladeTableLoc$Freq) %>% group_by(Country) %>% 
  mutate("N_over500"=sum(y), 
         "Observed_over500"=y,
         "Expected_over500"=round(N_over500*Prop)) %>% 
  select(-y)
#compare500Loc

#Chi-squared tests for oversample 500 Countries
list500_4 <- split(compare500Loc,compare500Loc$Country, drop=TRUE)
country500pVals <- as.data.frame(matrix(nrow=length(list500_4),ncol=3))
colnames(country500pVals) <- c("Country","p-Value","DF")
for (i in 1:length(list500_4)){
  temp <- as.data.frame(list500_4[i])
  colnames(temp) <- c("Clade","Country","Original","Prop","N_over500","Observed_over500","Expected_over500")
  temp <- temp %>% filter(Expected_over500!=0)
  if (nrow(temp)>=4){
    originalChiLoc500 <- chisq.test(temp$Observed_over500,correct=TRUE,p=temp$Expected_over500,rescale.p=TRUE)
    #Store chi squared tests in data frame
    #print(originalChiLoc500)
    temp <- data.frame(lapply(temp, as.character), stringsAsFactors=FALSE)
    country500pVals[i,1] <- temp[1,2]
    country500pVals[i,2] <- originalChiLoc500$p.value
    country500pVals[i,3] <- originalChiLoc500$parameter
  }
}
country500pVals <- country500pVals %>% drop_na
country500pVals
```