samovar-basic.Rmd

---
title: "Basic samovar usage"
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{samovar-basic}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

```{r, include = FALSE}
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  fig.width = 10
)

library(samovaR)
library(tidyverse)
```

# Download data

Download and vizualize data from GMrepo
```{r API}
teatree <- GMrepo_type2data(number_to_process = 1500)
viz_composition(teatree, type = "tile", interactive = F, top = 10)
```

# Preprocessing

## Filter data

```{r filter}
tealeaves <- teatree %>%
  teatree_trim(treshhold_species = 3, 
               treshhold_samples = 3, 
               treshhold_amount = 10^(-3))

```

## Normalizing

If you build teatree by your own, do rescaling stage when building via `teatree$rescale()` or assigning `teatree$min_value` and `teatree$max_value` is required

Good approximation to normal distribution is required for `glm` generating methods

```{r normalize}
teabag <- tealeaves %>%
  tealeaves_pack()
```

## Clustering

Do hierarchical (deprecated) or aggregating clustering of species

Remember: if you want to re-filter, it is better to re-do welding stage to avoid crashes in future!

```{r cluster}
concotion <- teabag %>%
  teabag_brew(min_cluster_size = 40, max_cluster_size = 150)
```

## Build samovar

```{r build}
data.samovar <- concotion %>%
  concotion_pour(probability_calculation = "simple")
```

## Generate data

```{r generate}
new_data <- data.samovar %>%
  samovar_boil(N = 100, avoid_zero_generations = T)
```

The generated data:
```{r generate_viz}
viz_composition(new_data, reord_samples = "hcl",type = "tile", interactive = F, top = 10)
```