wildman_index_bleed_april_2017.Rnw

\documentclass[ignorenonframetext,aspectratio=169]{beamer}
\usepackage{fontspec}
\setmainfont{FreeSerif}
\setsansfont{FreeSans}
\setmonofont{FreeMono}
\usepackage{url}
\usepackage{fancyhdr}
\usepackage{graphicx}
\usepackage[bf]{caption}
\usepackage{rotating}
\usepackage{wrapfig}
\usepackage{fancybox}
\usepackage{booktabs}
% \usepackage{multirow}
\usepackage{acronym}
\usepackage{qrcode}
\usepackage[backend=biber,hyperref=true,style=authoryear-comp,doi=false,url=false,eprint=false]{biblatex}
\addbibresource{references.bib}
% \usepackage[nomargin,inline,draft]{fixme}
% \newcommand{\DLA}[1]{\textcolor{red}{\fxnote{DLA: #1}}}
% \usepackage[hyperfigures,bookmarks,colorlinks,citecolor=black,filecolor=black,linkcolor=black,urlcolor=black]{hyperref}
\usepackage{texshade}
\usepackage{tikz}
\usepackage{nameref}
\usepackage{zref-xr,zref-user}
\renewcommand*{\bibfont}{\tiny}
\renewcommand*{\bibfont}{\fontsize{6pt}{6pt}\selectfont
}
\newrobustcmd*{\footlessfullcite}{\AtNextCite{\renewbibmacro{title}{}\renewbibmacro{in:}{}}\footfullcite}
\newrobustcmd*{\footlesscitetext}{\AtNextCite{\renewbibmacro{title}{}\renewbibmacro{in:}{}}\footcitetext}

% The textpos package is necessary to position textblocks at arbitary 
% places on the page.  Use showboxes option to show outlines of textboxes.
% \usepackage[absolute]{textpos}
\usepackage[absolute,overlay]{textpos}
\usepackage{mathtools,cancel}

% default to frame footnotes
\let\oldfootnote\footnote
\renewcommand\footnote[1][]{\oldfootnote[frame,#1]}

% include numbered bibliography items
\setbeamertemplate{bibliography item}{\insertbiblabel}


\renewcommand{\CancelColor}{\color{red}} %change cancel color to red
\newenvironment{digression}[1]{\begin{textblock*}{64mm}(0.2\textwidth,0.2\textheight)%
    \begin{block}{#1}}{%
\end{block}\end{textblock*}}


\usepackage{multirow}
\usepackage{array}

\mode<presentation>{ 
  \usetheme{CambridgeUS}
  % http://identitystandards.illinois.edu/graphicstandardsmanual/generalguidelines/colors.html
  \definecolor{ilboldblue}{HTML}{002058}
  \definecolor{ilboldorange}{HTML}{E87722}
  \definecolor{ilblue}{HTML}{606EB2}
  \definecolor{ilorange}{HTML}{D45D00}
  \setbeamercolor{alerted text}{fg=ilboldorange}
  \setbeamercolor*{palette primary}{fg=ilblue,bg=ilorange}
  \setbeamercolor*{palette secondary}{fg=ilblue!20!white,bg=ilorange}
  \setbeamercolor*{palette tertiary}{bg=ilblue,fg=ilorange}
  \setbeamercolor*{palette quaternary}{fg=ilblue,bg=ilorange}
  
  \setbeamercolor*{sidebar}{fg=ilorange,bg=ilboldblue}
  
  \setbeamercolor*{palette sidebar primary}{fg=ilblue!10!white,bg=ilorange}
  \setbeamercolor*{palette sidebar secondary}{fg=ilorange}
  \setbeamercolor*{palette sidebar tertiary}{fg=ilblue}
  \setbeamercolor*{palette sidebar quaternary}{fg=ilorange}
  
  % \setbeamercolor*{titlelike}{parent=palette primary}
  \setbeamercolor{titlelike}{parent=palette primary,fg=ilboldblue,bg=ilorange}
  \setbeamercolor{frametitle}{fg=ilboldorange!10!white,bg=ilblue!80!white}
  \setbeamercolor{frametitle right}{fg=ilboldblue,bg=ilorange}
  \setbeamercolor{block body}{bg=ilblue!10!white}
  \setbeamercolor{block title}{bg=ilorange!10!white}
  \setbeamertemplate{blocks}[rounded][shadow=false]
  \setbeamercolor*{separation line}{}
  \setbeamercolor*{fine separation line}{}
  \setbeamercovered{transparent}  
  \logo{\begin{tikzpicture}% Pale figure
      {\node[opacity=0.5]{\IfFileExists{figures/igb_wordmark.pdf}{\includegraphics[width=2cm,height=0.5cm,keepaspectratio]{figures/igb_wordmark}}{}%
        };}%
    \end{tikzpicture}}
}

% remove navigation symbols
\setbeamertemplate{navigation symbols}{}

\title[SeqSwitch]{Index sequence switching and Read Bleeding}
\author[Don Armstrong]{Don L. Armstrong}
\institute[IGB]{Institute for Genomic Biology, Computing Genomes 
  for Reproductive Health, University of Illinois, Urbana-Champaign}

\begin{document}

<<load.libraries,echo=FALSE,results="hide",warning=FALSE,message=FALSE,error=FALSE,cache=FALSE>>=
opts_chunk$set(dev="cairo_pdf",out.width="\\textwidth",out.height="0.8\\textheight",out.extra="keepaspectratio")
#opts_chunk$set(cache=TRUE, autodep=TRUE)
options(device = function(file, width = 8, height = 7, ...) {
  cairo_pdf(tempfile(), width = width, height = height, ...)
})
options(digits=2)
library("data.table")
library("ggplot2")
library("reshape2")
library("grid")
library("xtable")

@ 

\IfFileExists{figures/relevant_xkcd.png}{\frame[plain]{\centering \includegraphics[width=\textwidth,height=0.8\textheight,keepaspectratio]{figures/relevant_xkcd.png}

    \url{https://xkcd.com/1566/}}

\frame[plain]{\titlepage
  \begin{center}
    Code and slides are here: 
    
    \qrcode[padding]{https://github.com/uiuc-cgm/index-bleed-lab-meeting-april2017}
    
    \url{https://github.com/uiuc-cgm/index-bleed-lab-meeting-april2017}
  \end{center}
 }


\frame[plain]{
  \begin{center}
    \includegraphics[width=\textwidth,height=0.8\textheight,keepaspectratio]{index_switching_spread/paper_frontpage}
    
    \footlessfullcite{Sinha.ea2017:IndexSwitchingCausesSpreading-Of-Signal}
\end{center}
}

\begin{frame}{The Problem}
  \begin{columns}
    \column{0.6\textwidth}
    \includegraphics[width=\columnwidth,height=0.8\textheight,keepaspectratio]{index_switching_spread/figure_3a}
    \column{0.4\textwidth}
    \begin{itemize}
    \item High expression of \textit{Lgj} in row N703 column S513 has
      a curious pattern of expression in adjacent wells
    \end{itemize}    
  \end{columns}
\end{frame}

\section{Methods}
\begin{frame}{Method: Isolation}
  \begin{columns}
    \column{0.6\textwidth}
    \includegraphics[width=\columnwidth,height=0.8\textheight,keepaspectratio]{index_switching_spread/figure_2a}
    \column{0.4\textwidth}
    \begin{itemize}
    \item C57BL/6J mouse hematopoetic stem cells (mHSC) and mouse fetal heart cells (mFSC)
    \item “Highly purified” single cell sorted into skirted 96 well
      plates (A) with 4 μL lysis buffer
    \end{itemize}
  \end{columns}
\end{frame}

\begin{frame}{Method: Conversion}
  \begin{columns}
    \column{0.6\textwidth}
    \includegraphics[width=\columnwidth,height=0.8\textheight,keepaspectratio]{index_switching_spread/figure_2b}
    \column{0.4\textwidth}
    \begin{itemize}
    \item Lysate was converted to cDNA and amplified
    \item Quantified using fragment analyzer (B)
    \item Normalized to (0.05 ng/μL - 0.16 ng/μL) and consolidated
    \end{itemize}
  \end{columns}
\end{frame}

\begin{frame}{Method: Library prep and cleanup}
  \begin{columns}
    \column{0.5\textwidth}
    \includegraphics[width=\columnwidth,height=0.8\textheight,keepaspectratio]{index_switching_spread/figure_2c}
    \column{0.5\textwidth}
    \begin{itemize}
    \item Normalized to (0.05 ng/μL - 0.16 ng/μL) and consolidated
    \item Library preparation using nextera XT kit
    \item Cleaned up and size-selected using AMPure XP beads (excludes
      ≤ 500bp)
    \item Sequenced on a HiSeq 4000 or NextSeq 500
    \item demultiplexed using bcl2fastq, processed with skewer, and
      mapped using OLego
    \end{itemize}
  \end{columns}
\end{frame}

\begin{frame}{Methods: Size selection of fragments}
    \includegraphics[width=\columnwidth,height=0.8\textheight,keepaspectratio]{index_switching_spread/figure_S2}
\end{frame}

\section{Results}
\subsection{Signal Spreading}

\begin{frame}{Results: Signal spreading}
  \begin{columns}
    \column{0.5\textwidth}
    \includegraphics[width=\columnwidth,height=0.8\textheight,keepaspectratio]{index_switching_spread/figure_3a}
    \column{0.5\textwidth}
    \begin{itemize}
    \item High expression of \textit{Lgj} in row N703 column S513 has
      a curious pattern of expression in adjacent wells
    \item Were originally trying to use PCA to identify subpopulations
      of cells
    \item Each putative subpopulation mapped back to a single row or
      column in the plate
    \end{itemize}    
  \end{columns}
\end{frame}

\begin{frame}{Results: Signal spreading}
  \begin{columns}
    \column{0.5\textwidth}
    \includegraphics[width=\columnwidth,height=0.8\textheight,keepaspectratio]{index_switching_spread/figure_3b}
    \column{0.5\textwidth}
    \begin{itemize}
    \item High expression of \textit{Vip} in row N712 column S515
    \item Cells were randomly placed
    \item Must be artifact or systemic error
    \end{itemize}    
  \end{columns}
\end{frame}

\begin{frame}{Results: Signal spreading}
  \begin{columns}
    \column{0.5\textwidth}
    \includegraphics[width=\columnwidth,height=0.8\textheight,keepaspectratio]{index_switching_spread/figure_3c}
    \column{0.5\textwidth}
    \begin{itemize}
    \item High expression of \textit{Bmper} in row N718 column S508
    \item Identification of putative subpopulations were therefore not
      biologically meaningful
    \end{itemize}    
  \end{columns}
\end{frame}

\subsection{No RNA, still signal}

\begin{frame}{Exp 1 setup}
  \begin{columns}
    \column{0.5\textwidth}
    \includegraphics[width=\columnwidth,height=0.8\textheight,keepaspectratio]{index_switching_spread/figure_4}
    \column{0.5\textwidth}
    \begin{itemize}
    \item No index primers or RNA in 1-12
    \item No RNA in O24 and P24.
    \item B are free adapter/primer concentrations
    \end{itemize}
  \end{columns}
\end{frame}

\begin{frame}{Exp 1 results}
  \begin{columns}
    \column{0.5\textwidth}
    \includegraphics[width=\columnwidth,height=0.8\textheight,keepaspectratio]{index_switching_spread/figure_5}
    \column{0.5\textwidth}
    \begin{itemize}
    \item A is mapped reads in linear
    \item B is mapped reads in log scale
    \end{itemize}
  \end{columns}
\end{frame}

\begin{frame}{Exp 1 results summarized}
  \begin{columns}
    \column{0.6\textwidth}
    \includegraphics[width=\columnwidth,height=0.8\textheight,keepaspectratio]{index_switching_spread/figure_6}
    \column{0.4\textwidth}
    \begin{itemize}
    \item O24 and P24 had 1/10 of the reads of 12-24, even though
      there was no RNA
    \item Very little reads assigned to 1-12 where there were no index
      primers or RNA
    \item Contaminated reads likely due to free index primers
    \end{itemize}
  \end{columns}
\end{frame}

\subsection{Hypothesis?}

\begin{frame}{Barcode preparation?}
  \begin{columns}
    \column{0.5\textwidth}
    \includegraphics[width=\columnwidth,height=0.8\textheight,keepaspectratio]{index_switching_spread/figure_1}
    \column{0.5\textwidth}
    \begin{itemize}
    \item Library pool with residual free index primers gets mixed
    \item Free index primer anneals and gets extended to form new library molecule with switched index
    \item Separates from template and seeds a nanowell on the flow
      cell to generate a cluster
    \end{itemize}
  \end{columns}
\end{frame} 

\subsection{Tritrate in free index primers}

\begin{frame}{Exp 2 setup}
  \begin{columns}
    \column{0.5\textwidth}
    \includegraphics[width=\columnwidth,height=0.8\textheight,keepaspectratio]{index_switching_spread/figure_7}
    \column{0.5\textwidth}
    \begin{itemize}
    \item No RNA in 1-12 or O24 and P24
    \item Index primers added to 1, 3
    \item Index primers added to 1 (300 nM), 3 (10 nM) , 5 (1 nM)
    \end{itemize}
  \end{columns}
\end{frame}

\begin{frame}{Exp 2 results}
  \begin{columns}
    \column{0.5\textwidth}
    \includegraphics[width=\columnwidth,height=0.8\textheight,keepaspectratio]{index_switching_spread/figure_8}
    \column{0.5\textwidth}
    \begin{itemize}
    \item False signals only in columns (1,3) where index primers were added
    \end{itemize}
  \end{columns}
\end{frame}

\begin{frame}{Exp 2 results on NextSeq 500}
  \begin{columns}
    \column{0.5\textwidth}
    \includegraphics[width=\columnwidth,height=0.8\textheight,keepaspectratio]{index_switching_spread/figure_9}
    \column{0.5\textwidth}
    \begin{itemize}
    \item When run on NextSeq 500 significantly fewer redas mapping to
      wells in columns 1 and 3
    \end{itemize}
  \end{columns}
\end{frame}

\begin{frame}{Exp 2 results on different HiSeq 4000}
  \begin{columns}
    \column{0.5\textwidth}
    \includegraphics[width=\columnwidth,height=0.8\textheight,keepaspectratio]{index_switching_spread/figure_10}
    \column{0.5\textwidth}
    \begin{itemize}
    \item Index primers added to 1 (300 nM), 3 (10 nM) , 5 (1 nM)
    \item Reads assigned were directly proportional to the amount of
      spiked in primer
    \end{itemize}
  \end{columns}
\end{frame}

\begin{frame}{Exp 2 results on Quake lab NextSeq 500}
  \begin{columns}
    \column{0.5\textwidth}
    \includegraphics[width=\columnwidth,height=0.8\textheight,keepaspectratio]{index_switching_spread/figure_11}
    \column{0.5\textwidth}
    \begin{itemize}
    \item Index primers added to 1 (300 nM), 3 (10 nM) , 5 (1 nM)
    \item No large scale index switching
    \end{itemize}
  \end{columns}
\end{frame}

\begin{frame}{Exp 1 results on NextSeq 500}
  \begin{columns}
    \column{0.5\textwidth}
    \includegraphics[width=\columnwidth,height=0.4\textheight,keepaspectratio]{index_switching_spread/figure_12_6}
    
    \includegraphics[width=\columnwidth,height=0.4\textheight,keepaspectratio]{index_switching_spread/figure_12_5}
    \column{0.5\textwidth}
    \begin{itemize}
    \item No large scale index switching (top)
    \item Index switching on HiSeq 4000 (bottom)
    \end{itemize}
  \end{columns}
\end{frame}

\section{Discussion}

\begin{frame}{Summary}
  \begin{itemize}
  \item HiSeq 4000 has 5-7\% of reads in each sample may possibly
    belong to other samples in the library pool
  \item Since HiSeq 3000 and HiSeq X Ten share the same chemistry, it
    is possible that index switching is also occurring at a similar
    rate
  \item In their original analysis, all of the subpopulations were
    from signal spreading
  \end{itemize}
\end{frame}

\begin{frame}{What caused it?}
  \begin{itemize}
  \item Likely the isothermal amplification of library pool molecules,
    which only allows for a single library switch event, not two.
  \item Not limited to single-cell; potentially affects all library pools
  \item Just look for nextera forward and reverse index combinations
    which should not have reads assigned to them
  \item Very hard to detect cross-contamination in single index libraries
  \item The use of PhiX may acquire many of the free index primers,
    but because PhiX is non-indexed, clusters from PhiX are dark
  \item People are switching to an indexed PhiX to examine the amount
    of index switching which is occuring
  \end{itemize}
\end{frame}


\begin{frame}{What to do about it?}
  \begin{itemize}
  \item Avoid barcode sets with overlapping pairs and use paired-end
    reads
  \item Developing cleanup approaches which remove free primers, such
    as uracil-DNA glycosylase or PAGE purification of library pools
  \item<2-> Give up
  \end{itemize}
\end{frame}



\section*{References}

\begin{frame}[plain]{References}
  \begin{center}
    \mbox{}\vspace{-\baselineskip}
    \printbibliography[heading=none]
  \end{center}
\end{frame}

\end{document}