playing with MSA plots to allow filtering of positions, arghhh

scripts/plotting/get_plotting_dfs.R

@@ -110,9 +110,16 @@ merged_df3_comp = all_plot_dfs[[4]]
 ####################################################################
 
 #source(paste0(plot_script_path, "logo_data.R"))
-
 #s1 = c("\nSuccessfully sourced logo_data.R")
 #cat(s1)
+# input data is merged_df3
+# so repurposed it into a function so params can be passed instead to generate
+# data required for plotting.
+# Moved "logo_data.R" to redundant/
+
+source(paste0(plot_script_path, "logo_data_msa.R"))
+s1 = c("\nSuccessfully sourced logo_data_msa.R")
+cat(s1)
 
 ####################################################################
 #                        Data for DM OM Plots: Long format dfs
@@ -173,4 +180,4 @@ rm(c1
    , vars0
    , vars1
    , vars2
-   , vars3)
+   , vars3)

scripts/plotting/logo_data_msa.R

@@ -0,0 +1,49 @@
+#=================================================
+#         Data for Logo MSA plots
+#=================================================
+cat("\n=========================================="
+    , "\nLogo MSA Plots Data: ALL params"
+    , "\n=========================================")
+
+#msa1 = read.csv("/home/tanu/git/Data/cycloserine/output/gid_msa.csv", header = F)
+#head(msa1)
+#msa_seq= msa1$V1
+#head(msa_seq)
+
+#msa2 = read.csv("/home/tanu/git/Data/cycloserine/input/gid.1fasta", header = F)
+#head(msa2)
+#wt_seq = msa2$V1
+#head(wt_seq)
+
+# BOTH WORK
+#LogoPlotMSA(msa_seq, wt_seq)
+#LogoPlotMSA(msa1, msa2)
+#####################################
+
+#================
+# MSA file: muts
+#================
+
+in_filename_msa = paste0(tolower(gene), "_msa.csv") 
+infile_msa = paste0(outdir, "/", in_filename_msa)
+cat("\nInput file for MSA plots: ", infile_msa, "\n")
+
+msa1 = read.csv(infile_msa, header = F)
+head(msa1)
+cat("\nLength of MSA:", nrow(msa1))
+msa_seq = msa1$V1
+head(msa_seq)
+
+#================
+# fasta file: wt 
+#================
+  
+in_filename_fasta = paste0(tolower(gene), ".1fasta") 
+infile_fasta = paste0(indir, "/", in_filename_fasta)
+cat("\nInput fasta file for WT: ", infile_fasta, "\n")  
+
+msa2 = read.csv(infile_fasta, header = F)
+head(msa2)
+cat("\nLength of WT fasta:", nrow(msa2))
+wt_seq = msa2$V1
+head(wt_seq)