documentation update

export(STGmarkerFinder)

export(getDAcells)

export(getDAregion)

export(plotCellLabel)

export(plotCellScore)

export(plotDAsite)

export(runSTG)

export(updateDAcells)

import(RANN)

import(ggplot2)

import(Seurat)

import(cowplot)

import(ggplot2)

import(glmnet)

import(reticulate)

import(scales)

import(tclust)

importFrom(caret,createFolds)

#'

#' @param X matrix, normalized expression matrix of all cells in the dataset, genes are in rows,

#' rownames must be gene names

#' @param cell.idx result "da.cell.idx" from the output of function getDAcells

#' @param da.region.label result "cluster.res" from the output of function getDAregion

#' @param da.regions output from the function getDAregion()

#' @param da.regions.to.run numeric (vector), which DA regions to run the marker finder,

#' default is to run all regions

#' @param lambda numeric, regularization parameter that weights the number of selected genes,

#' @export

STGmarkerFinder <- function(

  X, cell.idx, da.region.label,

  X, da.regions,

  da.regions.to.run = NULL,

  lambda = 1.2, n.runs = 5, return.model = F,

  python.use = "/usr/bin/python", GPU = ""

  X.py <- r_to_py(as.matrix(X))

  # get DA regions to run

  n.da <- length(unique(da.region.label)) - 1

  n.da <- length(unique(da.regions$da.region.label)) - 1

  if(is.null(da.regions.to.run)){

    da.regions.to.run <- c(1:n.da)

  }

  # create DA label vector

  n.cells <- ncol(X)

  da.label <- rep(0, n.cells)

  da.label[cell.idx] <- da.region.label

  da.label <- da.regions$da.region.label

  # run model for each da region

## Introduction

DA-seq is a method to detect cell subpopulations with differential abundance between single cell RNA-seq (scRNA-seq) datasets from different samples, described in the preprint, "Detecting regions of differential abundance between scRNA-Seq datasets" available [here](https://www.biorxiv.org/content/10.1101/711929v2). Given a low dimensional transformation, for example principal component analysis (PCA), of the merged gene expression matrices from different samples (cell states, condition, etc.), DA-seq first computes a score vector for each cell to represent the DA behavior in the neighborhood to select cells in the most DA areas; then groups these cells into distinct DA regions.

This repository contains the DA-seq package.

[This](https://github.com/KlugerLab/DAseq) repository contains the DA-seq package.

## R Dependencies

Required packages: RANN, tclust, ggplot2, cowplot, RColorBrewer, scales, reticulate

<h2 class="hasAnchor">

<a href="#introduction" class="anchor"></a>Introduction</h2>

<p>DA-seq is a method to detect cell subpopulations with differential abundance between single cell RNA-seq (scRNA-seq) datasets from different samples, described in the preprint, “Detecting regions of differential abundance between scRNA-Seq datasets” available <a href="https://www.biorxiv.org/content/10.1101/711929v2">here</a>. Given a low dimensional transformation, for example principal component analysis (PCA), of the merged gene expression matrices from different samples (cell states, condition, etc.), DA-seq first computes a score vector for each cell to represent the DA behavior in the neighborhood to select cells in the most DA areas; then groups these cells into distinct DA regions.</p>

<p>This repository contains the DA-seq package.</p>

<p><a href="https://github.com/KlugerLab/DAseq">This</a> repository contains the DA-seq package.</p>

</div>

<div id="r-dependencies" class="section level2">

<h2 class="hasAnchor">

<meta property="og:title" content="DA-seq Step 1 &amp; Step 2: select DA cells — getDAcells" />

<meta property="og:description" content="Step 1: compute a multiscale score measure for each cell of its k-nearest-neighborhood for

multiple values of k.

Step 2: train a logistic regression classifier based on the multiscale score measure and retain cells

that may reside in DA regions." />

    <div class="ref-description">

    <p>Step 1: compute a multiscale score measure for each cell of its k-nearest-neighborhood for

multiple values of k.

Step 2: train a logistic regression classifier based on the multiscale score measure and retain cells

that may reside in DA regions.</p>

    </div>

    <pre class="usage"><span class='fu'>getDAcells</span>(<span class='no'>X</span>, <span class='no'>cell.labels</span>, <span class='no'>labels.1</span>, <span class='no'>labels.2</span>, <span class='no'>k.vector</span>, <span class='kw'>k.folds</span> <span class='kw'>=</span> <span class='fl'>10</span>,

  <span class='kw'>n.runs</span> <span class='kw'>=</span> <span class='fl'>10</span>, <span class='kw'>pred.thres</span> <span class='kw'>=</span> <span class='fu'><a href='https://rdrr.io/r/base/c.html'>c</a></span>(<span class='fl'>0.05</span>, <span class='fl'>0.95</span>), <span class='kw'>do.plot</span> <span class='kw'>=</span> <span class='no'>T</span>,

  <span class='kw'>plot.embedding</span> <span class='kw'>=</span> <span class='kw'>NULL</span>, <span class='kw'>size</span> <span class='kw'>=</span> <span class='fl'>0.5</span>, <span class='kw'>python.use</span> <span class='kw'>=</span> <span class='st'>"/usr/bin/python"</span>,

  <span class='kw'>GPU</span> <span class='kw'>=</span> <span class='st'>""</span>)</pre>

    <pre class="usage"><span class='fu'>getDAcells</span>(<span class='no'>X</span>, <span class='no'>cell.labels</span>, <span class='no'>labels.1</span>, <span class='no'>labels.2</span>, <span class='kw'>k.vector</span> <span class='kw'>=</span> <span class='kw'>NULL</span>,

  <span class='kw'>k.folds</span> <span class='kw'>=</span> <span class='fl'>10</span>, <span class='kw'>n.runs</span> <span class='kw'>=</span> <span class='fl'>10</span>, <span class='kw'>es.patience</span> <span class='kw'>=</span> <span class='fl'>10</span>, <span class='kw'>k.smooth</span> <span class='kw'>=</span> <span class='kw'>NULL</span>,

  <span class='kw'>pred.thres</span> <span class='kw'>=</span> <span class='fu'><a href='https://rdrr.io/r/base/c.html'>c</a></span>(<span class='fl'>0.05</span>, <span class='fl'>0.95</span>), <span class='kw'>do.plot</span> <span class='kw'>=</span> <span class='no'>T</span>, <span class='kw'>plot.embedding</span> <span class='kw'>=</span> <span class='kw'>NULL</span>,

  <span class='kw'>size</span> <span class='kw'>=</span> <span class='fl'>0.5</span>, <span class='kw'>python.use</span> <span class='kw'>=</span> <span class='st'>"/usr/bin/python"</span>, <span class='kw'>GPU</span> <span class='kw'>=</span> <span class='st'>""</span>)</pre>

    <h2 class="hasAnchor" id="arguments"><a class="anchor" href="#arguments"></a>Arguments</h2>

    <table class="ref-arguments">

    <colgroup><col class="name" /><col class="desc" /></colgroup>

    <tr>

      <th>X</th>

      <td><p>size N-by-p matrix, input merged dataset of interest after dimension reduction</p></td>

      <td><p>size N-by-p matrix, input merged dataset of interest after dimension reduction.</p></td>

    </tr>

    <tr>

      <th>cell.labels</th>

      <td><p>size N vector, labels for each input cell</p></td>

      <td><p>size N character vector, labels for each input cell</p></td>

    </tr>

    <tr>

      <th>labels.1</th>

      <td><p>vector, label name(s) that represent condition 1</p></td>

      <td><p>character vector, label name(s) that represent condition 1</p></td>

    </tr>

    <tr>

      <th>labels.2</th>

      <td><p>vector, label name(s) that represent condition 2</p></td>

      <td><p>character vector, label name(s) that represent condition 2</p></td>

    </tr>

    <tr>

      <th>k.vector</th>

      <td><p>integer, number of times to run the neural network to get the predictions, default 10</p></td>

    </tr>

    <tr>

      <th>pred.thres</th>

      <td><p>length-2 vector, top and bottom threshold on the predictions from the

logistic classification, default c(0.05,0.95)</p></td>

    </tr>

    <tr>

      <th>do.plot</th>

      <td><p>a logical value to indicate whether to return ggplot objects showing the results,

default True</p></td>

    </tr>

    <tr>

      <th>plot.embedding</th>

      <td><p>size N-by-2 matrix, 2D embedding for the cells</p></td>

    </tr>

    <tr>

      <th>size</th>

      <td><p>cell size to use in the plot, default 0.5</p></td>

    </tr>

    <tr>

      <th>python.use</th>

      <td><p>character string, the Python to use, default "/usr/bin/python"</p></td>

      <th>es.patience</th>

      <td><p>integer, patience parameter used in the EarlyStopping procedure, default 10</p></td>

    </tr>

    <tr>

      <th>GPU</th>

      <td><p>which GPU to use, default '', using CPU</p></td>

      <th>k.smooth</th>

      <td><p>integer, number of nearest neighbors used to smooth the prediction scores,

default 1<!-- % of cells} --></p>

<p>pred.threslength-2 vector, top and bottom threshold on the predictions from the

logistic classification, default c(0.05,0.95)</p>

<p>do.plota logical value to indicate whether to return ggplot objects showing the results,

default True</p>

<p>plot.embeddingsize N-by-2 matrix, 2D embedding for the cells</p>

<p>sizecell size to use in the plot, default 0.5</p>

<p>python.usecharacter string, the Python to use, default "/usr/bin/python"</p>

<p>GPUwhich GPU to use, default '', using CPU</p></td>

    </tr>

    </table>

    <h2 class="hasAnchor" id="value"><a class="anchor" href="#value"></a>Value</h2>

    <p>a list of results</p><dl'>

  <dt>da.ratio</dt><dd><p>score vector for each cell</p></dd>

  <dt>da.pred</dt><dd><p>(mean) prediction from the neural network</p></dd>

  <dt>da.cell.id</dt><dd><p>index for DA cells</p></dd>

  <dt>pred.plot</dt><dd><p>ggplot object showing the predictions of logistic regression on plot.embedding</p></dd>

  <dt>da.cells.plot</dt><dd><p>ggplot object highlighting cells of da.cell.idx on plot.embedding</p></dd>

</dl>

  </div>

  <div class="col-md-3 hidden-xs hidden-sm" id="pkgdown-sidebar">