taxo_scorer_user.Rmd

---
title: "Taxo Scorer (user version)"
---

#Lines having to be adapted to data

##Desired n° of initial candidates 
```{r}
top_initial <- 50
```

##Desired n° of final candidate(s) 
```{r}
top_final <- 5
```

##Taxonomical info of your sample(s)
```{r}
family <- "Papaveraceae"
genus <- "Glaucium"
species <- "Glaucium grandiflorum"
```

##Input
```{r}
# Path to initial annotation result
input_ISDB_path <- "https://osf.io/5uabh/download"

# Path to metadata with structures/biosources information
Metadata_path <- "~/190402_DNP_TAXcof_CF.tsv"
```

##Output
```{r}
# Path to weigthed annotation result
output_weighed_ISDB_path <- file(
  "~/Data/Glaucium_ISDB_DNP_results_top5_taxo_scored.tsv",
  encoding = "UTF-8"
)
```

#Requested packages
```{r}
library(data.table)
library(readr)
library(dplyr)
library(splitstackshape)
library(stringr)
library(purrr)
```

#Loading the files
```{r}
ISDBout <- read_delim(
  input_ISDB_path,
  delim = "\t",
  escape_double = FALSE,
  trim_ws = TRUE
)
Metadata <- read_delim(
  Metadata_path,
  delim = "\t",
  escape_double = FALSE,
  trim_ws = TRUE
)
```

#Preliminary cleaning
Splitting columns
```{r}
ISDBout <- ISDBout %>%
  as.data.table() %>%
  cSplit("SMILES_DNP", "|") %>%
  cSplit("InChIKey_DNP", "|") %>%
  cSplit("Biological_Source_DNP", "|") %>%
  cSplit("CRC_Number_DNP", "|") %>%
  cSplit("Molecular_Formula_DNP", "|") %>%
  cSplit("Compound_Types_DNP", "|") %>%
  cSplit("Molecule_Name_DNP", "|") %>%
  cSplit("Biological_Use_DNP", "|") %>%
  cSplit("Spectral_Score_DNP", "|") %>%
  cSplit("IS_libname", "|") %>%
  cSplit("IS_match_mzerror", "|")
```

Melting the data
```{r}
ISDBout_melted <- melt(setDT(ISDBout),
  id = ,
  measure = patterns(
    "^SMILES_DNP",
    "^InChIKey_DNP",
    "^Biological_Source_DNP",
    "^CRC_Number_DNP",
    "^Molecular_Formula_DNP",
    "^Compound_Types_DNP",
    "^Molecule_Name_DNP",
    "^Biological_Use_DNP",
    "^Spectral_Score_DNP",
    "^IS_libname",
    "^IS_match_mzerror"
  ),
  value.name = c(
    "SMILES_DNP",
    "InChIKey_DNP",
    "Biological_Source_DNP",
    "CRC_Number_DNP",
    "Molecular_Formula_DNP",
    "Compound_Types_DNP",
    "Molecule_Name_DNP",
    "Biological_Use_DNP",
    "Spectral_Score_DNP",
    "IS_libname",
    "IS_match_mzerror"
  ),
  variable.name = "Rank"
)
```

Generating ShortIK
```{r}
tempIK <- str_split_fixed(ISDBout_melted$InChIKey_DNP, "-", 3)
ISDBout_melted$"Short_IK_ISDB" <- tempIK[, 1]
```

Adding taxonomical information
```{r}
ISDBout_melted$Family_Bio <- family
ISDBout_melted$Genus_Bio <- genus
ISDBout_melted$Species_Bio <- species
```

Harmonizing output
```{r}
ISDBout <- ISDBout_melted %>%
  mutate_at(vars(matches("Rank")), list(as.numeric)) %>%
  select(
    ID_query = `cluster index`,
    Component_Index = componentindex,
    Molecular_Formula_DNP,
    Molecule_Name_DNP,
    Smiles_DNP = SMILES_DNP,
    Short_IK_query = Short_IK_ISDB,
    Rank_query = Rank,
    Score_query = Spectral_Score_DNP,
    Family_Bio,
    Genus_Bio,
    Species_Bio
  ) %>%
  data.table()
ISDBout <- ISDBout[ISDBout$Short_IK_query != "", ]
```

Keeping only the desired candidates
```{r}
ISDBout_top <- ISDBout %>%
  filter(Rank_query <= top_initial)
```

#Metadata
```{r}
Metadata_cleaned <- Metadata %>%
  select(
    Short_IK_DNP,
    Kingdom,
    Phylum,
    Class,
    Order,
    Family,
    Genus,
    Species,
    Subspecies,
    Kingdom_cf,
    Superclass_cf,
    Class_cf,
    Subclass_cf,
    Parent_Level_1_cf,
    Parent_Level_2_cf,
    Parent_Level_3_cf,
    Parent_Level_4_cf,
    Parent_Level_5_cf,
    Parent_Level_6_cf
  )
```

Adding metadata
```{r}
ISDBout_merged_taxed <- left_join(ISDBout_top, Metadata_cleaned, by = c("Short_IK_query" = "Short_IK_DNP"), match = "all")
```

#Function
Loading
```{r}
source("./manuscript_results/2_taxonomically_informed_scoring/taxo_functions.R")
```

Applying
```{r}
ISDBout_weighted <- TaxoWeighter_UseR(0.81, 1.62, 2.55, ISDBout_merged_taxed)
```

Keeping only desired top candidates
```{r}
ISDB_top5_weighted <- ISDBout_weighted %>%
  group_by(ID_query) %>%
  filter(RANK_FINAL <= top_final)
```

Outputing relevant information to Cytoscape friendly format
```{r}
ISDB2Cytoscape <- ISDB_top5_weighted %>%
  group_by(ID_query) %>%
  distinct(Short_IK_query, .keep_all = TRUE) %>%
  summarise(
    Score = paste(Score_query, collapse = "|"),
    Short_IK = paste(Short_IK_query, collapse = "|"),
    Smiles_DNP = paste(Smiles_DNP, collapse = "|"),
    Molecule_Name_DNP = paste(Molecule_Name_DNP, collapse = "|"),
    Molecular_Formula_DNP = paste(Molecular_Formula_DNP, collapse = "|"),
    Family = paste(Family, collapse = "|"),
    Genus = paste(Genus, collapse = "|"),
    Species = paste(Species, collapse = "|"),
    Subspecies = paste(Subspecies, collapse = "|"),
    Superclass_cf = paste(Superclass_cf, collapse = "|"),
    Class_cf = paste(Class_cf, collapse = "|"),
    Subclass_cf = paste(Subclass_cf, collapse = "|"),
    Parent_Level_1_cf = paste(Parent_Level_1_cf, collapse = "|"),
    Parent_Level_2_cf = paste(Parent_Level_2_cf, collapse = "|"),
    Parent_Level_3_cf = paste(Parent_Level_3_cf, collapse = "|"),
    Parent_Level_4_cf = paste(Parent_Level_4_cf, collapse = "|"),
    Parent_Level_5_cf = paste(Parent_Level_5_cf, collapse = "|"),
    Parent_Level_6_cf = paste(Parent_Level_6_cf, collapse = "|"),
    Pondered_Score_query = paste(Pondered_Score_query, collapse = "|"),
    Rank = paste(RANK_BEFORE, collapse = "|")
  )
```

Removing NA in Smiles for ChemViz plugin
```{r}
ISDB2Cytoscape$Smiles_DNP <- str_replace_all(ISDB2Cytoscape$Smiles_DNP, "NA", "")
```

#Exporting
```{r}
write.table(ISDB2Cytoscape, file = output_weighed_ISDB_path, row.names = FALSE, quote = FALSE, sep = "\t")
```