Available via license: CC BY 4.0
Content may be subject to copyright.
Exploring the single-cell
RNA-seq analysis
landscape with the scRNA-
tools database
Luke Zappia (1, 2)
Belinda Phipson (1)
Alicia Oshlack (1, 2)
1 Bioinformatics, Murdoch Children's Research Institute; 2 School of Biosciences,
University of Melbourne
Abstract
As single-cell RNA-sequencing (scRNA-seq) datasets have become more widespread the
number of tools designed to analyse these data has dramatically increased. Navigating
the vast sea of tools now available is becoming increasingly challenging for researchers.
In order to better facilitate selection of appropriate analysis tools we have been
cataloguing and curating new analysis tools, as they become available, in the scRNA-
tools database (www.scRNA-tools.org). Our database collects a range of information on
each scRNA-seq analysis tool and categorises them according to the analysis tasks they
perform. Exploration of this database gives insights into the areas of rapid development
of analysis methods for scRNA-seq data. We see that many tools are developed to
perform tasks specific to scRNA-seq analysis, particularly clustering and ordering of
cells. We also find that the scRNA-seq community embraces an open-source approach,
with most tools available under open-source licenses and preprints being extensively
used as a means to describe methods. The scRNA-tools database provides a valuable
resource for researchers embarking on scRNA-seq analysis and as a record of the growth
of the field over time.
.CC-BY 4.0 International licensea
certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under
The copyright holder for this preprint (which was notthis version posted October 20, 2017. ; https://doi.org/10.1101/206573doi: bioRxiv preprint
Keywords
Introduction
Single-cell RNA-sequencing (scRNA-seq) has rapidly gained traction as an effective tool
for interrogating the transcriptome at the resolution of individual cells. Since the first
protocols were published in 20091 the number of cells profiled in individual scRNA-seq
experiments has increased exponentially, outstripping Moore’s Law2. This new kind of
transcriptomic data brings a demand for new analysis methods. Not only does the scale
of scRNA-seq datasets vastly outstrip bulk experiments but there are also a variety of
challenges unique to the single-cell context3. Specifically, scRNA-seq data is extremely
sparse (there is no expression measured for many genes in most cells), it can technical
artefacts such as low-quality cells or differences between sequencing batches and the
scientific questions of interest are often different to those asked of bulk RNA-seq
datasets. For example many bulk RNA-seq datasets are generated to detect differentially
expressed genes through a designed experiment while many scRNA-seq experiments aim
to identify or classify cell types.
The bioinformatics community has embraced this new type of data, designing a plethora
of methods for the analysis of scRNA-seq data. As such, keeping up with the current state
of scRNA-seq analysis is now a significant challenge as the field is presented with a huge
number of choices for approaching an analysis. Since September 2016 we have collated
and categorised scRNA-seq analysis tools as they have become available. This database is
being continually updated and is publicly available at www.scRNA-tools.org. In order to
help researchers navigate the analysis jungle we discuss the stages of scRNA-seq analysis
and their relationship to tools and categories in the scRNA-tools database. Through the
analysis of this database we show trends in not only the analysis applications these
methods address but how they are published, licensed and the platforms they use. Based
on this database we gain insight into the state of analysis tools in this rapidly developing
field.
Overview of the scRNA-tools database
The scRNA-tools database contains information on software tools specifically designed
for the analysis of scRNA-seq data. For a tool to be eligible to be included in the database
.CC-BY 4.0 International licensea
certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under
The copyright holder for this preprint (which was notthis version posted October 20, 2017. ; https://doi.org/10.1101/206573doi: bioRxiv preprint
it must be available for download and public use. This can be from a software package
repository (such as Bioconductor4, CRAN or PyPI), a code sharing website such as
GitHub or directly from a private website. Various details of the tools are recorded such
as the programming language or platform they use, details of any related publication,
links to the source code and the associated software license. Tools are also categorised
according to the analysis tasks they are able to perform. Most tools are added after a
preprint or publication becomes available but some have been added after being
mentioned on social media or in similar collections such as Sean Davis' awesome-single-
cell page (https://github.com/seandavi/awesome-single-cell).
Information about tools is displayed on a publicly available website at www.scRNA-
tools.org. This website provides a profile for each tool, with links to publications and
code repositories, as well as an index by analysis category. We also provide an interactive
table that allows users to filter and sort tools to find those most relevant to their needs. A
final page shows live and up-to-date version of some of the analysis presented below.
Anyone wishing to contribute to the database can do so by submitting an issue to the
project GitHub page (https://github.com/Oshlack/scRNA-tools).
When the database was first constructed there were 70 scRNA-seq analysis tools
available, representing work in the field during the three years from the first published
tool in November 2013 (SAMstrt5) up to September 2016. In the year since then over 70
new tools have been added (Figure 1A). The doubling of the number of available tools in
such a short time demonstrates the booming interest in scRNA-seq and its maturation
from a technique requiring custom-built equipment with specialised protocols to a
commercially available product.
.CC-BY 4.0 International licensea
certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under
The copyright holder for this preprint (which was notthis version posted October 20, 2017. ; https://doi.org/10.1101/206573doi: bioRxiv preprint
Figure 1 (A) Number of tools in the scRNA-tools database over time. Since the scRNA-
seq tools database was started in September 2016 more than 70 new tools have been
released. (B) Publication status of tools in the scRNA-tools database. Over half of the
.CC-BY 4.0 International licensea
certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under
The copyright holder for this preprint (which was notthis version posted October 20, 2017. ; https://doi.org/10.1101/206573doi: bioRxiv preprint
tools in the full database have been published in peer-revewied journals while another
third are available as preprints. (C) When stratified by the date tools were added to the
database we see that the majority of tools added before October 2016 are published,
while the majority of newer tools are available as preprints. The number of
unpublished tools has stayed consistent at around 10 percent. (D) The majority of tools
are available using either the R or Python programming languages. (E) Most tools are
released under a standard open-source software license, with variants of the GNU
Public License (GPL) being the most common. However licenses could not be found for
a large proportion of tools.
Publication status
Most tools have been added to the scRNA-tools database after coming to our attention in
a paper describing their method and use. Of all the tools in the database about half have
been published in peer-reviewed journals and another third are described in preprint
articles, typically on the bioRxiv preprint server (Figure 1B). Tools can be split into those
that were available when the database was created and those that have been added since.
We can see that the majority of older tools have been published while more recent tools
are more likely to only be available as preprints (Figure 1C). This is a good
demonstration of the delay imposed by the traditional publication process. By publishing
preprints and releasing software via repositories such as GitHub scRNA-seq tool
developers make their methods available to the community much earlier, allowing them
to be used for analysis and their methods improved prior to formal publication.
Platforms and licensing
Developers of scRNA-seq analysis tools have choices to make about what platforms they
use to create their tools, how they make them available to the community and whether
they share the source code. We find that the most commonly used platform for creating
scRNA-seq analysis tools is the R statistical programming language, with many tools
made available through the Bioconductor or CRAN repositories (Figure 1D). Python is
the second most popular language, followed by MATLAB, a commercially available
product, and the lower-level C++. The use of R and Python is consistent with their
popularity across a range of data science fields. In particular the popularity of R reflects
its history as the language of choice for the analysis of bulk RNA-seq datasets and a
range of other biological data types.
The majority of tools in the scRNA-tools database have chosen to take an open-source
approach, making their code available under permissive licenses (Figure 1E). We feel this
reflects the general underlying sentiment and willingness of the bioinformatics
.CC-BY 4.0 International licensea
certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under
The copyright holder for this preprint (which was notthis version posted October 20, 2017. ; https://doi.org/10.1101/206573doi: bioRxiv preprint
community to share and build upon the work of others. Variations of the GNU Public
License (GPL) are the most common, covering almost half of tools. This license allows
free use, modification and distribution of source code, but also has a "copyleft" nature
which requires any derivatives to disclose their source code and use the same license.
The MIT license is the second most popular which also allows use of code for any
purpose but without any restrictions on distribution or licensing. The appropriate license
could not be identified for almost a fifth of tools. This is problematic as fellow developers
must assume that source code cannot be reused, potentially limiting the usefulness of the
methods in those tools. Tool owners are strongly encouraged to clearly display their
license in source code and documentation to provide certainty to the community as to
how their work can be reused.
Categories of scRNA-seq analysis
As has been described in previous reviews a standard scRNA-seq analysis consists of
several tasks which can be completed using various tools6. In the scRNA-tools database
we categorise tools based on the analysis tasks they perform. Here we group these tasks
into four broad phases of analysis: data acquisition, data cleaning, cell assignment and
gene identification (Figure 2). The data acquisition phase (Phase 1) takes the raw
nucleotide sequences from the sequencing experiment and returns a matrix describing
the expression of each gene (rows) in each cell (columns). This phase consists of tasks
common to bulk RNA-seq experiments, such as alignment to a reference genome or
transcriptome and quantification of expression, but is often extended to handle Unique
Molecular Identifiers (UMIs). Once an expression matrix has been obtained it is vital to
make sure the resulting data is of high enough quality. In the data cleaning phase (Phase
2) quality control of cells is performed as well as filtering of uninformative genes.
Additional tasks may be performed to normalise the data or impute missing values.
Exploratory data analysis tasks are often performed in this phase, such as viewing the
datasets in reduced dimensions to look for underlying structure.
The high-quality expression matrix is the focus of the next phases of analysis. In Phase 3
cells are assigned, either to discrete groups via clustering or along a continuous
trajectory from one cell type to another. As high-quality reference datasets become
available it will also become feasible to classify cell directly into different cell types. Once
cells have been assigned attention turns to interpreting what those assignments mean.
.CC-BY 4.0 International licensea
certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under
The copyright holder for this preprint (which was notthis version posted October 20, 2017. ; https://doi.org/10.1101/206573doi: bioRxiv preprint
Identifying interesting genes (Phase 4), such as those that are differentially expressed
across groups, marker genes expressed in a single group or genes that change expression
along a trajectory, is the typical way to do this. The biological significance of those genes
can then be interpreted to give meaning to the experiment, either by investigating the
genes themselves or by getting a higher level view through techniques such as gene set
testing.
.CC-BY 4.0 International licensea
certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under
The copyright holder for this preprint (which was notthis version posted October 20, 2017. ; https://doi.org/10.1101/206573doi: bioRxiv preprint
Figure 2 Phases of the scRNA-seq analysis process. In Phase 1 (data acquisition) raw
sequencing reads are converted into a gene by cell expression matrix. For many
protocols this requires the alignment of genes to a reference genome and the
assignment and de-duplication of Unique Molecular Identifiers (UMIs). The data is
then cleaned (Phase 2) to remove low-quality cells and uninformative genes, resulting
in a high-quality dataset for further analysis. The data can also be normalised and
missing values imputed during this phase. Phase 3 assigns cells, either in a discrete
manner to known (classification) or unknown (clustering) groups or to a position on a
continuous trajectory. Interesting genes (eg. differentially expressed, markers, specific
patterns of expression) are then identified to explain these groups or trajectories (Phase
4).
While there are other approaches that could be taken to analyse scRNA-seq data these
phases represent the most common path from raw sequencing reads to biological insight.
Descriptions of the categories in the scRNA-tools database are given in Table 1, along
with the associated analysis phases.
Table 1 Descriptions of categories for tools in the scRNA-tools database
Phase
Category
Description
Phase 1
Alignment
Alignment of sequencing reads to a reference
Phase 1
Assembly
Tools that perform assembly of scRNA-seq reads
Phase 1
UMIs
Processing of Unique Molecular Identifiers
Phase 1
Quantification
Quantification of expression from reads, including
handling unique molecular identifiers
Phase 2
Quality Control
Removal of low-quality cells
Phase 2
Gene Filtering
Removal of lowly expressed or otherwise uninformative
genes
Phase 2
Imputation
Estimation of expression where zeros have been
observed
Phase 2
Normalisation
Removal of unwanted variation that may affect results
Phase 2
Cell Cycle
Assignment or correction of stages of the cell cycle, or
other uses of cell cycle genes, or genes associated with
similar processes
Phase 3
Classification
Assignment of cell types based on a reference dataset
Phase 3
Clustering
Unsupervised grouping of cells based on expression
profiles
.CC-BY 4.0 International licensea
certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under
The copyright holder for this preprint (which was notthis version posted October 20, 2017. ; https://doi.org/10.1101/206573doi: bioRxiv preprint
Phase 3
Ordering
Ordering of cells along a trajectory
Phase 3
Rare Cells
Identification of rare cell populations
Phase 3
Stem Cells
Identification of cells with stem-like characteristics
Phase 4
Differential
Expression
Testing of differential expression across groups of cells
Phase 4
Expression
Patterns
Detection of genes that change expression across a
trajectory
Phase 4
Gene Networks
Identification of co-regulated gene networks
Phase 4
Gene Sets
Testing for over representation or other uses of
annotated gene sets
Phase 4
Marker Genes
Identification or use of genes that mark cell populations
Multiple
Dimensionality
Reduction
Projection of cells into a lower dimensional space
Multiple
Interactive
Tools with an interactive component or a graphical user
interface
Multiple
Variable Genes
Identifcation or use of highly (or lowly) variable genes
Multiple
Visualisation
Functions for visualising some aspect of scRNA-seq
data or analysis
Other
Allele Specific
Detection of allele-specific expression
Other
Alternative
Splicing
Detection of alternative splicing
Other
Haplotypes
Use or assignment of haplotypes
Other
Immune
Assignment of receptor sequences and immune cell
clonality
Other
Integration
Combining of scRNA-seq datasets or integration with
other single-cell data types
Other
Modality
Identification or use of modality in gene expression
Other
Simulation
Generation of synthetic scRNA-seq datasets
Other
Transformation
Transformation between expression levels and some
other measure
.CC-BY 4.0 International licensea
certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under
The copyright holder for this preprint (which was notthis version posted October 20, 2017. ; https://doi.org/10.1101/206573doi: bioRxiv preprint
Other
Variants
Detection or use of nucleotide variants
Trends in scRNA-seq analysis tasks
Each of the tools in the database is assigned to one or more analysis categories. We
investigated these categories in further detail to give insight into the trends in scRNA-seq
analysis. Figure 3A shows the frequency of tools performing each of the analysis tasks.
Visualisation is the most commonly included task and is important across all stages of
analysis for exploring and displaying data and results. Tasks for assigning cells (ordering
and clustering) are the next most common. This has been the biggest area of
development in single-cell analysis with clustering tools such as Seurat11, SC312 and
BackSPIN13 being used to identify cell types in a sample and trajectory analysis tools (for
example Monocle14, Wishbone15 and DPT16) being used to investigate how genes change
across developmental processes. These areas reflect the new opportunities for analysis
provided by single-cell data that are not possible with bulk RNA-seq experiments.
.CC-BY 4.0 International licensea
certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under
The copyright holder for this preprint (which was notthis version posted October 20, 2017. ; https://doi.org/10.1101/206573doi: bioRxiv preprint
Figure 3 (A) Categories of tools in the scRNA-tools database. Each tool can be assigned
to multiple categories based on the tasks it can complete. Categories associated with
multiple analysis phases (visualisation, dimensionality reduction) are among the most
.CC-BY 4.0 International licensea
certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under
The copyright holder for this preprint (which was notthis version posted October 20, 2017. ; https://doi.org/10.1101/206573doi: bioRxiv preprint
common, as are categories associated with the cell assignment phase (ordering,
clustering). (B) Changes in analysis categories over time, comparing tools added before
and after October 2016. There have been significant increases in the percentage of tools
associated with visualisation, dimensionality reduction, quantification and simulation.
Categories including expression patterns, pseudotime and interactivity have seen
relative decreases. (C) Changes in the percentage of tools associated with analysis
phases over time. The percentage of tools involved in the data acquisition and data
cleaning phases have increased, as have tools designed for alternative analysis tasks.
The gene identification phase has seen a relative decrease in the number of tools. (D)
The number of categories associated with each tools in the scRNA-tools database. The
majority of tools perform few tasks. (E) Most tools that complete many tasks are
relatively recent.
Dimensionality reduction is also a common task and has applications in visualisation
(via techniques such as t-SNE17), quality control and as a starting point for analysis.
Testing for differential expression (DE) is perhaps the most common analysis performed
on bulk RNA-seq datasets and it is also commonly applied by many scRNA-seq analysis
tools, typically to identify genes that are different in one cluster of cells compared to the
rest. However it should be noted that the DE testing applied by scRNA-seq tools is often
not as sophisticated as the rigorous statistical frameworks of tools developed for bulk
RNA-seq such as edgeR18, DESeq220 and limma21, often using simple statistical tests such
as the likelihood ratio test. While methods designed to test DE specifically in single-cell
datasets do exist (such as SCDE22, and scDD23) it is still unclear whether they improve on
methods that have been established for bulk data24.
To investigate how the focus of scRNA-seq tool development has changed over time we
again divided the scRNA-tools database into tools added before and after October 2016.
This allowed us to see which analysis tasks are more common in recently published tools.
We looked at the percentage of tools in each time period that performed tasks in the
different analysis categories (Figure 3B). Some categories show little change in the
proportion of tools that perform while other areas have changed significantly.
Specifically, both visualisation and dimensionality reduction are more commonly
addressed by recent tools. The UMIs category has also seen a big increase recently as
UMI based protocols have become commonly used and tools designed to handle the
extra processing steps required have been developed (UMI-tools28, umis29, zUMIs30).
Simulation is a valuable technique for developing, testing and validating scRNA-seq
tools. More packages are now including their simulation functions and some tools have
been developed for the specific purpose of generating realistic synthetic scRNA-seq
datasets (powsimR31, Splatter32). Classification of cells into known groups has also
.CC-BY 4.0 International licensea
certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under
The copyright holder for this preprint (which was notthis version posted October 20, 2017. ; https://doi.org/10.1101/206573doi: bioRxiv preprint
increased as reference datasets become available and more tools are identifying or
making use of co-regulated gene networks.
Some categories have seen a decrease in the proportion of tools they represent, most
strikingly testing for changes expression patterns along a trajectory. This is likely related
to the change in cell ordering analysis which is the focus of a lower percentage of tools
added after October 2016. The ordering of cells along a trajectory was one of the first
developments in scRNA-seq analysis and a decrease in the development of these tools
could indicate that researchers have moved on to other techniques or that use has
converged on a set of mature tools.
By grouping categories based on their associated analysis phases we see similar trends
over time (Figure 3C). We see increases in the percentage of tools performing tasks in
Phase 1 (quantification), Phase 2 (quality control and filtering), across multiple phases
(visualisation and dimensionality reduction) and alternative analysis tasks. In contrast
the percentage of tools that perform gene identification tasks (Phase 2) has decreased
and the percentage assigning cells (Phase 3) has remained steady. This too may indicate
a maturation of the analysis space as existing tools for performing standard scRNA-seq
analyses are deemed sufficient while there is still room for development in handling data
from new protocols and performing alternative analysis tasks.
Pipelines and toolboxes
While there are a considerable number of scRNA-seq tools that only perform a single
analysis task, many perform at least two (Figure 3D). Some tools (dropEst33, DrSeq234,
scPipe35) are preprocessing pipelines, taking raw sequencing reads and producing an
expression matrix. Others, such as Scanpy36, SCell37, Seurat, Monocle and scater38 can be
thought of as analysis toolboxes, able to complete a range of complex analyses starting
with a gene expression matrix. Most of the tools that complete many tasks are more
recent (Figure 3E). Being able to complete multiple tasks using a single tool can simplify
analysis as problems with converting between different data formats can be avoided,
however it is important to remember that it is difficult for a tool with many
functionalities to continue to represent the state of the art in all of them. Support for
common data formats, such as the recently released SingleCellExperiment object in R39,
provides another way for developers to allow easy use of their tools and users to build
custom workflows from specialised tools.
.CC-BY 4.0 International licensea
certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under
The copyright holder for this preprint (which was notthis version posted October 20, 2017. ; https://doi.org/10.1101/206573doi: bioRxiv preprint
Alternative analyses
Some tools perform analyses that lie outside the common tasks performed on scRNA-seq
data described above. Simulation is one alternative task that has already been mentioned
but there is also a group of tools designed to detect biological signals in scRNA-seq data
apart from changes in expression. For example alternative splicing (BRIE40, Outrigger41,
SingleSplice42), single nucleotide variants (SSrGE43) and allele-specific expression
(SCALE44). Reconstruction of immune cell receptors is another area that has received
considerable attention from tools such as BASIC45, TraCeR46 and TRAPeS47. While tools
that complete these tasks are unlikely to ever dominate scRNA-seq analysis it is likely
that we will see an increase in methods for tackling specialised analyses as researchers
continue to push the boundaries of what can be observed using scRNA-seq data.
Discussion and conclusions
Over the last year we have seen the number of number of software tools for analysing
single-cell RNA-seq data double, with more than 130 analysis tools now available. As
new tools have become available we have curated and catalogued them in the scRNA-
tools database where we record the analysis tasks that they can complete, along with
additional information such as any associated publications. By analysing this database
we have found that tool developers have focused much of their efforts on methods for
handling new problems specific to scRNA-seq data, in particular clustering cells into
groups or ordering them along a trajectory. We have also seen that the scRNA-seq
community is generally open and willing to share their methods which are often
described in preprints prior to peer-reviewed publication and released under permissive
open-source licenses for other researchers to re-use.
The next few years promise to continue to produce significant new developments in
scRNA-seq analysis. New tools will continue to be produced, becoming increasingly
sophisticated and aiming to address more of the questions made possible by scRNA-seq
data. We anticipate that some existing tools will continue to improve and expand their
functionality while others will cease to be updated and maintained. Detailed
benchmarking and comparisons will show how tools perform in different situations and
those that perform well, continue to be developed and provide a good user experience
will become preferred for standard analyses. As single-cell capture and sequencing
.CC-BY 4.0 International licensea
certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under
The copyright holder for this preprint (which was notthis version posted October 20, 2017. ; https://doi.org/10.1101/206573doi: bioRxiv preprint
technology continues to improve analysis tools will have to adapt to significantly larger
datasets (in the millions of cells) which may require specialised data structures and
algorithms. Methods for combining multiple scRNA-seq datasets as well as integration of
scRNA-seq data with other single-cell data types, such as DNA-seq, ATAC-seq or
methylation, with be another area of growth and projects such as the Human Cell Atlas48
will provide comprehensive cell type references which will open up new avenues for
analysis.
As the field expands the scRNA-tools database will continue to be updated. We hope that
it provides a resource for researchers to explore when approaching scRNA-seq analyses
as well as providing a record of the analysis landscape and how it changes over time.
Methods
Database
When new tools come to our attention they are added to the scRNA-tools database. DOIs
and publication dates are recorded for any associated publications. As preprints may be
frequently updated they are marked as a preprint instead of recording a date. The
platform used to build the tool, links to code repositories, associated licenses and a short
description are also recorded. Each tool is categorised according to the analysis tasks it
can perform, receiving a true or false for each category based on what is described in the
accompanying paper or documentation. We also record the date that each entry was
added to the database and the date that it was last updated.
Website
To build the website we start with the table described above as a CSV file which is
processed using an R script. The lists of packages available in the CRAN, Bioconductor
and PyPI software repositories are downloaded and matched with tools in the database.
For tools with peer-reviewed publications the number of citations they have received is
retrieved from the Crossref database (www.crossref.org) using the rcrossref package
(v0.7.0)49. JSON files describing the complete table, tools and categories are outputted
and used to populate the website.
The website consists of three main pages. The home page shows an interactive table with
the ability to sort, filter and download the database. The second page shows an entry for
.CC-BY 4.0 International licensea
certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under
The copyright holder for this preprint (which was notthis version posted October 20, 2017. ; https://doi.org/10.1101/206573doi: bioRxiv preprint
each tool, giving the description, details of publications, details of the software code and
license and the associated software categories. Badges are added to tools to provide
clearly visible details of any associated software or GitHub repositories. The final page
describes the categories, providing easy access to the tools associated with them.
Analysis
The most recent version of the scRNA-tools database was used for the analysis presented
in this paper. Data was manipulated in R using the dplyr package (v0.7.3)50 and plots
produced using the ggplot2 (v2.2.1)51 and cowplot (v0.8.0)52 packages.
Declarations
Ethics
Not applicable.
Availability of data and materials
The scRNA-tools databases is publicly accessible via the website at www.scRNA-
tools.org. Suggestions for additions, updates and improvements are warmly welcomed at
the associated GitHub repository (https://github.com/Oshlack/scRNA-tools). The code
and datasets used for the analysis in this paper are available from
https://github.com/Oshlack/scRNAtools-paper.
Competing interests
The authors declare no competing interests.
Funding
Luke Zappia is supported by an Australian Government Research Training Program
(RTP) Scholarship. Alicia Oshlack is supported through a National Health and Medical
Research Council Career Development Fellowship APP1126157. MCRI is supported by
the Victorian Government's Operational Infrastructure Support Program.
.CC-BY 4.0 International licensea
certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under
The copyright holder for this preprint (which was notthis version posted October 20, 2017. ; https://doi.org/10.1101/206573doi: bioRxiv preprint
Authors' contributions
Acknowledgements
We would like to acknowledge Sean Davis' work in managing the awesome-single-cell
page and producing a prototype of the script used to process the database. Daniel Wells
had the idea for recording software licenses and provided licenses for the tools in the
database at that time. Breon Schmidt designed a prototype of the scRNA-tools website
and answered many questions about HTML and Javascript. Our thanks also to Matt
Ritchie for his thoughts on early versions of the manuscript.
Additional files
References
1. Tang, F. et al. mRNA-Seq whole-transcriptome analysis of a single cell. Nature
methods 6, 377–382 (2009).
2. Svensson, V., Vento-Tormo, R. & Teichmann, S. A. Moore’s Law in Single Cell
Transcriptomics. (2017).
3. Stegle, O., Teichmann, S. A. & Marioni, J. C. Computational and analytical challenges
in single-cell transcriptomics. Nature Reviews Genetics 16, 133–145 (2015).
4. Huber, W. et al. Orchestrating high-throughput genomic analysis with Bioconductor.
Nature methods 12, 115–121 (2015).
5. Katayama, S., Töhönen, V., Linnarsson, S. & Kere, J. SAMstrt: statistical test for
differential expression in single-cell transcriptome with spike-in normalization.
Bioinformatics 29, 2943–2945 (2013).
6. Bacher, R. & Kendziorski, C. Design and computational analysis of single-cell RNA-
sequencing experiments. Genome biology 17, 63 (2016).
7. Wagner, A., Regev, A. & Yosef, N. Revealing the vectors of cellular identity with single-
cell genomics. Nature biotechnology 34, 1145–1160 (2016).
8. Miragaia, R. J., Teichmann, S. A. & Hagai, T. Single-cell insights into transcriptomic
diversity in immunity. Current Opinion in Systems Biology 5, 63–71 (2017).
9. Poirion, O. B., Zhu, X., Ching, T. & Garmire, L. Single-Cell Transcriptomics
Bioinformatics and Computational Challenges. Frontiers in genetics 7, (2016).
10. Rostom, R., Svensson, V., Teichmann, S. A. & Kar, G. Computational approaches for
interpreting scRNA-seq data. FEBS letters (2017). doi:10.1002/1873-3468.12684
.CC-BY 4.0 International licensea
certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under
The copyright holder for this preprint (which was notthis version posted October 20, 2017. ; https://doi.org/10.1101/206573doi: bioRxiv preprint
11. Satija, R., Farrell, J. A., Gennert, D., Schier, A. F. & Regev, A. Spatial reconstruction
of single-cell gene expression data. Nature biotechnology 33, 495–502 (2015).
12. Kiselev, V. Y. et al. SC3: consensus clustering of single-cell RNA-seq data. Nature
methods 14, 483–486 (2017).
13. Zeisel, A. et al. Brain structure. Cell types in the mouse cortex and hippocampus
revealed by single-cell RNA-seq. Science 347, 1138–1142 (2015).
14. Trapnell, C. et al. The dynamics and regulators of cell fate decisions are revealed by
pseudotemporal ordering of single cells. Nature biotechnology 32, 381–386 (2014).
15. Setty, M. et al. Wishbone identifies bifurcating developmental trajectories from
single-cell data. Nature biotechnology 34, 637–645 (2016).
16. Haghverdi, L., Büttner, M., Wolf, F. A., Buettner, F. & Theis, F. J. Diffusion
pseudotime robustly reconstructs lineage branching. Nature methods (2016).
doi:10.1038/nmeth.3971
17. Maaten, L. van der & Hinton, G. Visualizing Data using t-SNE. Journal of machine
learning research: JMLR 9, 2579–2605 (2008).
18. Robinson, M. D., McCarthy, D. J. & Smyth, G. K. edgeR: a Bioconductor package for
differential expression analysis of digital gene expression data. Bioinformatics 26, 139–
140 (2010).
19. McCarthy, D. J., Chen, Y. & Smyth, G. K. Differential expression analysis of
multifactor RNA-Seq experiments with respect to biological variation. Nucleic acids
research 40, 4288–4297 (2012).
20. Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and
dispersion for RNA-seq data with DESeq2. Genome biology 15, 550 (2014).
21. Ritchie, M. E. et al. limma powers differential expression analyses for RNA-
sequencing and microarray studies. Nucleic acids research 43, e47 (2015).
22. Kharchenko, P. V., Silberstein, L. & Scadden, D. T. Bayesian approach to single-cell
differential expression analysis. Nature methods 11, 740–742 (2014).
23. Korthauer, K. D. et al. A statistical approach for identifying differential distributions
in single-cell RNA-seq experiments. Genome biology 17, 222 (2016).
24. Soneson, C. & Robinson, M. D. Bias, Robustness And Scalability In Differential
Expression Analysis Of Single-Cell RNA-Seq Data. bioRxiv 143289 (2017).
doi:10.1101/143289
25. Jaakkola, M. K., Seyednasrollah, F., Mehmood, A. & Elo, L. L. Comparison of
methods to detect differentially expressed genes between single-cell populations.
Briefings in bioinformatics (2016). doi:10.1093/bib/bbw057
26. Miao, Z. & Zhang, X. Differential expression analyses for single-cell RNA-Seq: old
questions on new data. Quantitative Biology 4, 243–260 (2016).
27. Dal Molin, A., Baruzzo, G. & Di Camillo, B. Single-cell RNA-sequencing: assessment
of differential expression analysis methods. Frontiers in genetics 8, 62 (2017).
.CC-BY 4.0 International licensea
certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under
The copyright holder for this preprint (which was notthis version posted October 20, 2017. ; https://doi.org/10.1101/206573doi: bioRxiv preprint
28. Smith, T., Heger, A. & Sudbery, I. UMI-tools: modeling sequencing errors in Unique
Molecular Identifiers to improve quantification accuracy. Genome research 27, 491–499
(2017).
29. Svensson, V. et al. Power analysis of single-cell RNA-sequencing experiments.
Nature methods (2017). doi:10.1038/nmeth.4220
30. Parekh, S., Ziegenhain, C., Vieth, B., Enard, W. & Hellmann, I. zUMIs: A fast and
flexible pipeline to process RNA sequencing data with UMIs. bioRxiv 153940 (2017).
doi:10.1101/153940
31. Vieth, B., Ziegenhain, C., Parekh, S., Enard, W. & Hellmann, I. powsim: Power
analysis for bulk and single cell RNA-seq experiments. bioRxiv 117150 (2017).
doi:10.1101/117150
32. Zappia, L., Phipson, B. & Oshlack, A. Splatter: simulation of single-cell RNA
sequencing data. Genome biology 18, 174 (2017).
33. Petukhov, V. et al. Accurate estimation of molecular counts in droplet-based single-
cell RNA-seq experiments. bioRxiv 171496 (2017). doi:10.1101/171496
34. Zhao, C., Hu, S., Huo, X. & Zhang, Y. Dr.seq2: A quality control and analysis pipeline
for parallel single cell transcriptome and epigenome data. PloS one 12, e0180583 (2017).
35. Tian, L. et al. scPipe: a flexible data preprocessing pipeline for single-cell RNA-
sequencing data. bioRxiv 175927 (2017). doi:10.1101/175927
36. Alexander Wolf, F., Angerer, P. & Theis, F. J. Scanpy for analysis of large-scale
single-cell gene expression data. bioRxiv 174029 (2017). doi:10.1101/174029
37. Diaz, A. et al. SCell: integrated analysis of single-cell RNA-seq data. Bioinformatics
(2016). doi:10.1093/bioinformatics/btw201
38. McCarthy, D. J., Campbell, K. R., Lun, A. T. L. & Wills, Q. F. Scater: pre-processing,
quality control, normalization and visualization of single-cell RNA-seq data in R.
Bioinformatics 33, 1179–1186 (2017).
39. Lun, A. & Risso, D. SingleCellExperiment: S4 Classes for Single Cell Data. (2017).
40. Huang, Y. & Sanguinetti, G. BRIE: transcriptome-wide splicing quantification in
single cells. Genome biology 18, 123 (2017).
41. Song, Y. et al. Single-Cell Alternative Splicing Analysis with Expedition Reveals
Splicing Dynamics during Neuron Differentiation. Molecular cell (2017).
doi:10.1016/j.molcel.2017.06.003
42. Welch, J. D., Hu, Y. & Prins, J. F. Robust detection of alternative splicing in a
population of single cells. Nucleic acids research 44, e73 (2016).
43. Poirion, O. B., Zhu, X., Ching, T. & Garmire, L. X. Using Single Nucleotide Variations
in Cancer Single-Cell RNA-Seq Data for Subpopulation Identification and Genotype-
phenotype Linkage Analysis. bioRxiv 095810 (2016). doi:10.1101/095810
44. Jiang, Y., Zhang, N. R. & Li, M. SCALE: modeling allele-specific gene expression by
single-cell RNA sequencing. Genome biology 18, 74 (2017).
.CC-BY 4.0 International licensea
certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under
The copyright holder for this preprint (which was notthis version posted October 20, 2017. ; https://doi.org/10.1101/206573doi: bioRxiv preprint
45. Canzar, S., Neu, K. E., Tang, Q., Wilson, P. C. & Khan, A. A. BASIC: BCR assembly
from single cells. Bioinformatics 33, 425–427 (2017).
46. Stubbington, M. J. T. et al. T cell fate and clonality inference from single-cell
transcriptomes. Nature methods 13, 329–332 (2016).
47. Afik, S. et al. Targeted reconstruction of T cell receptor sequence from single cell
RNA-seq links CDR3 length to T cell differentiation state. Nucleic acids research (2017).
doi:10.1093/nar/gkx615
48. Regev, A. et al. The Human Cell Atlas. bioRxiv 121202 (2017). doi:10.1101/121202
49. Chamberlain, S., Boettiger, C., Hart, T. & Ram, K. rcrossref: Client for Various
’CrossRef’ ’APIs’. (2017).
50. Wickham, H., Francois, R., Henry, L. & Müller, K. dplyr: A Grammar of Data
Manipulation. (2017).
51. Wickham, H. ggplot2: Elegant Graphics for Data Analysis. (Springer New York,
2010).
52. Wilke, C. O. cowplot: Streamlined Plot Theme and Plot Annotations for ’ggplot2’.
(2017).
.CC-BY 4.0 International licensea
certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under
The copyright holder for this preprint (which was notthis version posted October 20, 2017. ; https://doi.org/10.1101/206573doi: bioRxiv preprint
