analysis of multiple beta-gal experiments.Rmd

---
title: "R Notebook"
output:
  html_document:
    df_print: paged
---

This code imports, analyzes, and plots data from a multiple replcicates of beta-galactosidase assays. This analysis is for experiments examining beta-galactosidase activity as a function of RpoS level.

Before running this code, you must have followed the directions and run the code in `analysis of single beta-gal experiment.Rmd` on every file you intend analyze here. 

```{r, message=FALSE}
library(readxl)
library(tidyverse)
source("betaGalAnalysisFunctions.R")
theme_set(theme_classic())

fileNames <- c("sampleData/6-6-18 LB32_LB73.xlsx","sampleData/6-6-18 LB32_LB73 fake data for testing.xlsx")

processedFiles <- map(fileNames, meanExpressionOneReplicate)

combinedData <- bind_rows(processedFiles)
```

```{r}
ggplot(combinedData) + geom_point(aes(x=RpoS, y=meanExpression)) + facet_grid(.~strain) 
```

Plot mean ± SE for each RpoS level
```{r}
dayMeans <- combinedData %>% 
              group_by(strain, RpoS) %>% 
              summarise(meanAcrossDays = mean(meanExpression), sdAcrossDays = sd(meanExpression)/sqrt(length(meanExpression))) %>%
              mutate(upperErrorBar = meanAcrossDays + sdAcrossDays, lowerErrorBar = meanAcrossDays - sdAcrossDays)

ggplot(dayMeans) + 
  geom_point(aes(x = RpoS, y = meanAcrossDays)) + 
  geom_errorbar(aes(x = RpoS, y = meanAcrossDays, ymax = upperErrorBar, ymin = lowerErrorBar)) + 
  facet_grid(.~strain)

  
```
##Sensitivity

This section is specific for analyzing the sensitivity of a promoter. It requires three RpoS levels.

This function expects three RpoS levels, so for this example I'll subset out some, since we have five.

```{r}
dataForSensitivity <- combinedData %>% 
                        filter(RpoS == 0 | RpoS == 19 | RpoS == 94) 
```


Calculate sensitivity of a promoter. 
```{r}
sensitivities <- dataForSensitivity %>% 
                  group_by(strain, date, experimenter) %>%
                  summarise(sensitivity = calculateSingleSens(RpoS, meanExpression))
```


Plot the data with a line to illustrate null expectation used to assess sensitivity. 


```{r}
segmentEnds <- dayMeans %>%
                filter(RpoS == 0 | RpoS == 94) %>%
                group_by(strain) %>%
                summarise(
                  RpoS_min = min(RpoS),
                  RpoS_max = max(RpoS),
                  expression_min = min(meanAcrossDays),
                  expression_max = max(meanAcrossDays)
                  )
  
```


If you designed your experiment with only three RpoS levels, delete the the `filter` step. Note that we have to go back a step to use use the `dayMeans` tibble.
```{r}
dayMeans %>% 
  filter(RpoS == 0 | RpoS == 19 | RpoS == 94) %>% 
  ggplot() + 
  geom_point(aes(x = RpoS, y = meanAcrossDays, size = 4), show.legend = FALSE) + 
  geom_errorbar(aes(x = RpoS, y = meanAcrossDays, ymax = upperErrorBar, ymin = lowerErrorBar)) + 
  facet_grid(.~strain) +
  geom_segment(data = segmentEnds, mapping = aes(x = RpoS_min, xend = RpoS_max, y = expression_min, yend = expression_max)) +
  labs(x = "% of wild-type RpoS", y = "LacZ expression (Miller Units)")

                 
```