'''

This program is meant to take the Trace files outputted from ebFRET and determine the lifetime of each state in a given HMM
before the state transitions to another state. At the moment this program generates files that can then be given to MEMLET
to determine the rate constants for each state in a given HMM, though.

'''
#Packages

library(tidyverse)

#Define global variables

path_var <- r"(C:\Users\clark\Desktop\Documents for Thesis and Manuscript\Analysis_9.9.2022\Lifetime_of_Initial_Sampling_Event_CognateNoTig\ebFRET\Lifetime_of_Initial_Sampling_Event_CognateNoTig)"
file_name <- "2stateHMM_Lifetime_of_Initial_Sampling_Event_CognateNoTig.dat"
setwd(path_var)
time_interval <- 0.04

#Load in the Full Report Trace outputs from ebFRET

file_location <- paste(path_var, file_name, sep = "\\")
TypeAll_traces_2states_ebFRET <- read.table(file_location, quote="\"", comment.char="")

#Name the columns for each dataframe

colnames(TypeAll_traces_2states_ebFRET) <- c("Trace", "Donor", "Acceptor", "FRET", "Viterbi_State", "Viterbi_Mean")


#Use the rle() function to determine the number of occurances in a row for each state in the 2-state HMM

Two_State_rle <- rle(TypeAll_traces_2states_ebFRET$Viterbi_State)
Two_State_rle <- do.call(rbind.data.frame, Two_State_rle)
Two_State_rle <- t(Two_State_rle)

row.names(Two_State_rle) <- NULL
colnames(Two_State_rle) <- c("Num_Occur", "State")
Two_State_rle <- as_tibble(Two_State_rle)

Two_State_rle <- Two_State_rle %>% mutate(Num_Occur = Num_Occur*time_interval)  #Convert number of frames to time in seconds


#Use dplyr to filter out the number of occurances in a row for each state in the 2-state HMM

Length_State1_2HMM <- filter(Two_State_rle, State == 1)
Length_State2_2HMM <- filter(Two_State_rle, State == 2)

write.table(Length_State1_2HMM$Num_Occur, "NearCognate_25nMTC_PRE6_SecondaryEvents_State1_2HMM.txt", row.names = FALSE, col.names = FALSE)
write.table(Length_State2_2HMM$Num_Occur, "NearCognate_25nMTC_PRE6_SecondaryEvents_State2_2HMM.txt", row.names = FALSE, col.names = FALSE)