---
title: "Synaptome Statistics: July 2017 Notes"
date: '`r Sys.Date()`'
output:
  html_document:
    fig_caption: yes
    fig_height: 5
    fig_width: 5
    highlight: pygments
    number_sections: yes
    theme: cerulean
    toc: yes
    toc_depth: 3
---
```{r knitOPTS, include=FALSE}
knitr::opts_chunk$set(cache = TRUE, dev = "png")
```

```{r render, eval=FALSE, echo=FALSE}
require(rmarkdown)
rm(list=ls()); 
rmarkdown::render("Notes201707.Rmd")
system("open Notes201707.html")
```

```{r setup,include=FALSE,results='asis',message=FALSE,warning=FALSE, echo = FALSE}
### Library calls here.
require(rmarkdown)
require(knitr)
suppressMessages(require(meda))
```

# Weiler14 Ex2R18C1

Putative synapse locations were obtained by Anish Simhal.  
Around each synapse across each channel the sum of pixel intensities within an 11x11x11 cube are calculated to produce a feature vector. 
These feature vectors are combined into a data matrix.  We replace the
DAPI channels with their first principal component.  The data matrix is
then z-scored before running it through [meda](https://github.com/neurodata-dev/meda). 

We did this for samples of size 1,000 and 10,000. 

## Sample size 1,000

### 1-d Heatmap
![](weiler1k/output1k/plots/2.png)

### Location plots
![](weiler1k/output1k/plots/raw5.png)

### Outliers as given by randomForest
![](weiler1k/output1k/plots/6.png)

### Correlation Matrix
![](weiler1k/output1k/plots/4.png)

### Cumulative Variance with Elbows
![](weiler1k/output1k/plots/1.png)

### Paired Hex-binned plot
![](weiler1k/output1k/plots/7.png)

### Hierarchical GMM Classifications
![](weiler1k/output1k/plots/3.png)

### Hierarchical GMM Dendrogram
![](weiler1k/output1k/plots/dend.png)

### Stacked Cluster Means plot
```{r stackM, include = FALSE, eval=FALSE}
suppressPackageStartupMessages(require(meda))
h <- readRDS("weiler1k/output1k/hmc.rds")
png("weiler1k/output1k/plots/stackMM.png", height = 1800, width = 800)
plot(stackM(h, ccol = h$ccol, centered = TRUE)) 
dev.off()
```
![](weiler1k/output1k/plots/stackM.png)

### Cluster Means
![](weiler1k/output1k/plots/clusterMeans.png)


## Sample size 10,000
### 1-d Heatmap
![](weiler10k/output10k/plots/2.png)

### Location plots
![](weiler10k/output10k/plots/raw5.png)

### Outliers as given by randomForest
![](weiler10k/output10k/plots/6.png)

### Correlation Matrix
![](weiler10k/output10k/plots/4.png)

### Cumulative Variance with Elbows
![](weiler10k/output10k/plots/1.png)

### Paired Hex-binned plot
![](weiler10k/output10k/plots/7.png)

### Hierarchical GMM Classifications
![](weiler10k/output10k/plots/3.png)

### Hierarchical GMM Dendrogram
![](weiler10k/output10k/plots/dend.png)

### Stacked Cluster Means plot
```{r stackM10k, include = FALSE, eval = FALSE}
suppressPackageStartupMessages(require(meda))
h <- readRDS("weiler10k/output10k/hmc.rds")
png("weiler10k/output10k/plots/stackMM.png", height = 1800, width = 800)
plot(stackM(h, ccol = h$ccol, centered = TRUE)) 
dev.off()
```
![](weiler10k/output10k/plots/stackM.png)

### Cluster Means
![](weiler10k/output10k/plots/clusterMeans.png)




--- 
---