add_phase.Rmd

---
title: "Add phases after EPMA analysis"
author: "YASUMOTO Atsushi"
date: "`r Sys.Date()`"
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{Add phases after EPMA analysis}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

<!-- © 2018 JAMSTEC -->

```{r setup, include = FALSE}
wd <- tempdir()
knitr::opts_chunk$set(
  root.dir = wd,
  fig.width = 5,
  fig.height = 5,
  collapse = TRUE,
  comment = "#>"
)
library(qntmap)
library(dplyr)
library(pipeR)
library(ggplot2)
```

# Intro

This document introduces a way to quantify X-ray map even if, in accident, 
there are some phases without spot analysis. 
For accurate analysis, I highly recommend careful observation prior to EPMA analysis, 
so that as less phases as possible being missed.

Please read ["Get started"](qntmap.html) before hand.

# Preparation

This section prepares package, files, and modify sample data.
An original sample dataset is same as that used in ["Get started"](qntmap.html).
However, the dataset will be modified to let a phase be not analyzed by EPMA spot analysis.
This is why "[Specify directories containing data]" and "[Load Data]" belongs to "[Preparation]" despite
they belongs to [Analysis section of "Basic usage"](basic#Analysis).

## Load package

```{r load-qntmap, eval = FALSE}
library(qntmap)
```

## Move to working directory

```{r setwd, eval = FALSE}
wd <- tempdir() # Arbitrary
setwd(wd)
```

## Copy example data

or any original data.

```{r copy-example-data}
file.copy(
  from = system.file('extdata', 'minimal', package = 'qntmap'),
  to = wd,
  recursive = TRUE
)
```

`TRUE` indicates files are successfully copied. 
Check if really files are copied by `dir()`.

```{r check-file-structure}
dir(file.path(wd, 'minimal'), recursive = TRUE, all.files = TRUE)
```

## Specify directories containing data


```{r specify-paths, include = FALSE}
dir_map <- file.path(wd, 'minimal/.map/1')
dir_qnt <- file.path(wd, 'minimal/.qnt')
```

``` r
dir_map <- 'minimal/.map/1'
dir_qnt <- 'minimal/.qnt'
```


## Load data

```{r read-data, message = FALSE}
xmap <- read_xmap(dir_map)
qnt <- read_qnt(dir_qnt)
```

## Check quantified data

In the original dataet, 20 spots are analyzed by EPMA on both olivine and quartz.

```{r summarize-qnt}
table(qnt$cnd$phase)
```

## Modify data

Let's delete quartz from dataset and reload.

After `read_qnt()` is performed, 
a file `phase_list0.csv` is created under working directory.
This file consists of 3 columns `id`, `phase`, and `use`.

Edit **only** `phase` and `use`. Do not edit `id`.
If certain phases have large compositional variations, fill `phase` column with
different names.
(e.g., Ol_Fe and Ol_Mg for Fe-rich olivine and Mg-rich olivine).
If certain analysis show bad results, fill `use` column with `FALSE`, otherwise `TRUE`.

To do it on R, execute codes below.
Be sure to specify a path to modified csv file to `phase_list` parameter of `read_qnt`.
Also, `renew` parameter of `read_qnt` must be `TRUE`.

```{r phase-list}
phase_list <- read.csv('phase_list0.csv')
phase_list$use[phase_list$phase == 'Qtz'] <- FALSE
write.csv(phase_list, file.path(dir_qnt, 'phase_list_no_qtz.csv'))
qnt <- read_qnt(dir_qnt, phase_list = file.path(dir_qnt, 'phase_list_no_qtz.csv'))
```

After editing, only olivine is said to be quantified.

```{r table-phase}
table(qnt$cnd$phase)
```

```{r plot-quantified-points, message = FALSE, echo = FALSE}
epma <- tidy_epma(qnt, xmap) %>>%
  filter(elm == elm[[1]]) %>>%
  select(x_px, y_px, phase)

plot(xmap, 'Si', interactive = FALSE, colors = "gray") +
  geom_point(
    aes(y_px, x_px, colour = phase),
    data = epma, inherit.aes = FALSE
  ) +
  guides(
    fill = guide_colourbar(barheight = grid::unit(1, "npc") - unit(10, "line"))
  )
```

# Analysis

## Cluster analysis

### Initialize cluster centers

```{r init-centers}
centers <- find_centers(xmap, qnt)
centers
```

### Compare the initial centroids with X-ray map plotted with heatmap

```{r plot-si}
plot(xmap, 'Si', interactive = FALSE)
```

In this case, an analysist will notice there is something other than olivine.
Let's mouse over the **interactive map** and one will see one of the coordinates of quartz (e.g., x = 18, y  = 28).
Keep it on your note.

### Add initial centers

#### on R

Add initial centers by using `add_centers`.

```{r add-centers}
centers <- add_centers(centers = centers, xmap = xmap, x = 18, y = 28, p = 'Qtz')
```

Then, Qtz is properly added to `centers`.

```{r print-centers}
print(centers)
```
#### on Spreadsheet program + R

- Open `centers_initial0.csv` in the current directory (try `getwd()` on R if unknown).
- Edit it
- Save it
- Reload on R by `centers <- read.csv('path to an updated csv file')`

### Cluster and quantify

As Qtz is not quantified during spot analysis,
calibration curves for Qtz is substituted by results of regression analysis
without considering different phases.
Try "[Quasi-calibrations]" in case phases not quantified have known and 
constant chemical compositions.

```{r quantify, fig.show = 'hide'}
cluster <- cluster_xmap(xmap, centers)
qmap <- quantify(xmap, qnt, cluster)
```

```{r summary}
summary(cluster)
summary(qmap)
```

#### Quasi-calibrations

In case phases without spot analysis have known chemical compositions,
their calibration curves can be calculated by 
specifying their chemical compositions in csv format.

For example, prepare a csv file, and specify it to `fix` parameter of `quantify()`.

```{r csv}
csv <- paste(
    'phase, oxide, wt',
    'Qtz,   SiO2,      100',
    sep = "\n"
  )
cat(csv)
```

```{r quantify2}
qmap2 <- quantify(xmap, qnt, cluster, fix = csv)
summary(qmap2)
```

Of course, `csv` parameter can be a path to a csv file, 
unlike the above example, which gave csv data the `csv` parameter.




```{r on-exit-unlink, include = FALSE}
unlink(c('centers0.csv', 'center_add.csv'))
```