-
Notifications
You must be signed in to change notification settings - Fork 8
/
Copy pathREADME.Rmd
215 lines (100 loc) · 19.7 KB
/
README.Rmd
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
---
output: github_document
---
<!-- README.md is generated from README.Rmd. Please edit that file -->
```{r setup, include = FALSE}
knitr::opts_chunk$set(
collapse = TRUE,
comment = "#>",
fig.path = "man/figures/README-",
out.width = "100%",
fig.align = "center"
)
```
# StructuRly 0.1.0
**StructuRly** is an **R** package containing a [shiny](https://shiny.rstudio.com) application to produce detailed and interactive graphs of the results of a Bayesian cluster analysis obtained with the most common population genetic software used to investigate population structure, such as *STRUCTURE* or *ADMIXTURE*. These software are widely used to infer the admixture ancestry of samples starting from genetic markers such as SNPs, AFLPs, RFLPs and microsatellites (such as SSRs). More generally, **StructuRly** can generate graphs from any file containing admixture information of each sample (encoded in percentages in a range from 0 to 1).
We developed **StructuRly** to provide researchers with detailed graphical outputs to interpret their statistical results through the use of software with a user-friendly interface, which can, therefore, be easily used by those who do not know a programming language. In fact, in a typical **StructuRly** output, the user will have the possibility to display information about the ID of each sample, the original membership assigned by the researcher to the sampled populations (or subpopulations) and the label of the sampling site, a variable, the latter, which is used in software for population analysis to support the data analysis algorithm. Furthermore, interactivity is a typical feature of **StructuRly** outputs, which allows the user to extrapolate even more information through a single chart.
However, this shiny application presents more different features to:
- support the statistical genetic analysis with necessary information about the molecular markers and diversity indices and through the calculation of the $P_{gen}$ (if you have haploid or diploid data) or the Hardy-Weinberg equilibrium for every locus. For the calculation of the Hardy-Weinberg equilibrium, the $p$-value of the $\chi^2$-test can be calculated for any level of ploidy (>= 2), while the exact $p$-value from the Monte Carlo test is currently available just for diploids (more details are available inside the [pegas](https://cran.r-project.org/web/packages/pegas/pegas.pdf) package manual);
- upload datasets with raw genetic data to analyze them through the principal coordinates analysis (MDS) and hierarchical cluster analysis algorithms, and view and download the dendrograms based on different distance matrices and linkage methods;
- produce and customize tables ready to be imported into the *STRUCTURE* software for the Bayesian analysis;
- import the results of the *STRUCTURE* and *ADMIXTURE* population analysis directly into **StructuRly** in different formats, without having to re-structure the dataset with other software (such as **R**);
- produce an interactive barplot and triangle plot, the most well-known *STRUCTURE* graphical outputs. Both graphs can show the admixture ancestry of the samples subdivided in a maximum of 20 different clusters;
- visually compare the partition obtained from the hierarchical cluster analysis and the one from the Bayesian (*STRUCTURE*) or maximum likelihood (*ADMIXTURE*) analysis through a confusion matrix and estimate an agreement value of the two partitions with two different agreement indices.
- visualize and download the *R* code used inside the shiny application to produce all the plots.
### Installation
You can install the released version of **StructuRly** from [GitHub](https://github.com) in *RStudio* with:
``` r
install.packages(pkgs = "devtools")
library(devtools)
install_github(repo = "nicocriscuolo/StructuRly", dependencies = TRUE)
```
Once the package is loaded and the dependencies installed, you can run the software in the default browser through the following functions:
``` r
library(StructuRly)
runStructuRly()
```
If you have trouble installing **StructuRly** you can follow the instructions present this [link](https://github.com/nicocriscuolo/StructuRly/blob/master/inst/Instructions_install_StructuRly.txt).
### System requirements
**StructuRly** works on macOS, Windows and Linux operative systems. Install the updated version of [R](https://www.r-project.org) (>= 3.5) and [RStudio](https://www.rstudio.com) and launch **StructuRly** on all types of browsers (Internet Explorer, Safari, Chrome, etc.). In its current version, it can also work locally and then offline. If you need any information about the usage of *STRUCTURE* or *ADMIXTURE* software (e. g. instructions to launch the software, preparation of input files and how to exports the outputs), please visit their websites at the following links:
- [STRUCTURE](https://web.stanford.edu/group/pritchardlab/structure.html)
- [ADMIXTURE](http://software.genetics.ucla.edu/admixture/)
Moreover, the user can launch the Terminal (to start an *ADMIXTURE* population analysis) or the *STRUCTURE* software directly from the user interface of **StructurRly** (this function is currently available for macOS and Linux users). To make this buttons work, both software must be installed on your computer.
**N. B.**: If you use a Linux based machine, to properly configure **R** and to install some **StructuRly** dependencies you may need specific Linux libraries to make these software work with this operative system. To install these libraries in **R** follow the instructions displayed inside the **R** console when you load the dependency packages.
### Online version
If you are not familiar with *R* or *RStudio* you can access to **StructuRly** directly from the web by using the following link: [https://nicocriscuolo1618.shinyapps.io/StructuRly/](https://nicocriscuolo1618.shinyapps.io/StructuRly/).
### Data input
**StructuRly** is divided into three different sections depending on the input file chosen. For any type of file, the **header of each variable is mandatory** and varies according to the type of variable that must be present in the input dataset. When you start a new session of **StructuRly**, if you change the uploaded file with a new one (inside the same section), to produce new outputs remember to re-define every time the **type of separator** (e. g. column, semi-column or tab) and to indicate if your data have **quotation marks**.
###### Data format
In the first section of **StructuRly**, you can import both **.txt** and **.csv** file. Since the second section also accepts the output file obtained after the population analysis performed with *ADMIXTURE*, here you can import also **.Q** format file and a **.fam** file (if the latter one is available).
In **StructuRly** you also have the possibility to export a table ready to be imported inside the *STRUCTURE* software. If you need detailed references about the structure of this dataset and how to perform the population analysis with *STUCTURE* you can find them this [link](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3665925/).
If you want to use your raw genetic data to produce an input table for the *ADMIXTURE* software, you have to convert your matrix in a **.ped** or **.bed** file. You can do that through the functionalities of the *PLINK* software, illustrated step by step at this [link](http://zzz.bwh.harvard.edu/plink/data.shtml). If you need more information about this last data formats, they are available [here](https://www.cog-genomics.org/plink2/formats#bed).
###### Download sample datasets
Examples of the **.txt**, **.csv**, **.Q** and **.fam** files that you can import into **StructuRly** are present at the following repository link: [Sample datasets](https://github.com/nicocriscuolo/StructuRly/tree/master/inst/Sample_dataset) (the .Q and the .fam files are obtained after an *ADMIXTURE* analysis with the sample files downloadable directly from the [ADMIXTURE](http://software.genetics.ucla.edu/admixture/) website).
To download the sample datasets from GitHub, right-click on the desired file and choose **Download linked file**. The sample datasets are available in pair of two files: one contains the raw genetic data and the other the results of the *STRUCTURE* analysis performed on such data. They have different format and information to describe different use-case scenario, in particular:
- **Sample1**: this datasets in **.txt** format contains random generated values of genetic triploid loci (with different names) in 500 samples, with a weight that ranges from 150 to 500 base-pairs. The additional information available are the Sample ID, the Population ID and Location ID (see Section 1);
- **Sample2**: this datasets in **.csv** format contain information related to diploid genetic loci of simple sequence repeats (SSR) sampled in 95 *Olea europaea* specimens in [Criscuolo et al., 2019](https://www.mdpi.com/2223-7747/8/9/297). They contain additional information about the Sample ID and the Population ID;
- **Sample3**: the last sample dataset is in **.Q** format and contains the results of the *ADMIXTURE* analysis on a genetic dataset available on the *ADMIXTURE* website. Moreover, the **.fam** file is available to add the Sample ID and the Population ID to the original dataset.
#### Section 1: Import raw genetic data
The input for this section can contain three optional variables present in the following order and whose header must be precisely the one shown below:
* **Sample_ID**: is the variable that contains the IDs of each sample so each name in this column will be different from the others (although it is good practice to use only numbers and letters, the IDs characters can also be separated by the following symbols: "_" and "-");
* **Pop_ID**: is a categorical variable identified by an integer that indicates the putative population defined by the user for each sample (e.g .: 1, 2, 3, etc.);
* **Loc_ID**: another categorical variable identified again by an integer that indicates the origin site of each sample; this variable, if present in the table produced with **StructuRly** and then imported into *STRUCTURE* for population analysis, can be used by the Bayesian algorithm as support to results elaboration.
The following variables present in the dataset to import in this section are mandatory and must contain numerical values relative to the types of markers used. Depending on the ploidy of the organism analyzed, there must be a number of columns for each locus equal to the number of alleles, in particular:
* for haploid organisms, each column must have a unique name of a locus coded with alpha-numeric characters, but it must not contain the dot symbol (e. g.: "Locus_1", "Locus_2", "UDO", "BRAC8792", etc.);
* for diploid and polyploid organisms the column header must contain a single name of the locus followed, this time, by the dot symbol (".") and the identification number of the allele, mandatorily starting from 1 (e.g .: "Locus_1.1," "Locus_1.2", "UDO.1", "UDO.2", etc.). Below there is an image that represents data stored in a spreadsheet that, once converted in **.txt** or **.csv** format, can be appropriately read by **StructuRly**:
**N. B.**: for the **Sample_ID**, **Pop_ID** and **Loc_ID** columns, avoid the usage of the name "NA" to indicate a name of a sample, of a putative population or a collection site, because **StructuRly** could recognize that characters as a missing value and the plot will not display the correct information. This also applies for the preparation of the input datasets for the Section 2.
###### Missing values
When you produce the file for this section of **StructuRly**, the missing values must be indicated only with the abbreviation **NA**. The cells of the reactive table (in the table panel named "Input table") that contain missing values will appear empty, while they are codified as **-9** in the table that can be produced and downloaded by **StructuRly** to be imported into *STRUCTURE*.
**N. B.**: if your data refer to diploid or polyploid organisms and you encounter a missing value in one or more of your samples in a specific locus, the **NA** value must be present for all the alleles of that locus;
#### Section 2: Import population analysis
Here the user can import a dataset obtained directly following the population analysis of his genetic data. The characteristics of this input file are not very different from the one to be imported in the previous section:
* the three optional variables (**Sample_ID**, **Pop_ID** and **Loc_ID**) can be inserted again, in this precise order, within the import file, with the only difference that in this case the categories of the variables **Pop_ID** and **Loc_ID** do not necessarily have to be represented by numbers, but also by characters.
* the other mandatory variables to be inserted must be those of the admixture proportion calculated by the population software mentioned above, and which will be equal in number to the number of clusters chosen by the user before executing the Bayesian analysis. Each of these variables must be identified by a header containing the letter "K" and the number of the relative cluster in sequence (e. g.: "K1", "K2", "K3", etc.), i. e. in the same order of the dataset exported by the software.
Below there is an example of this type of file structure. In this case the **Loc_ID** column is not present; in fact, the three information variable are not mandatory for the datasets to import in section 1. and 2.:
* If you have obtained the results of your population analysis with the *ADMIXTURE* software, there are two ways to proceed to prepare the dataset for **StructuRly**. From the analysis of a file in **.bed** or **.ped** format you will get a **.Q** format file that you can either import into **R** and then modify as you like, exporting it in **.txt** or **.csv** format and then import it into **StructuRly** (for example, after adding the columns identifying the name of the samples or the population) or you can import directly into **StructuRly** the **.Q** format file. This file only contains the variables with the values of the ancestry admixture: if you want to add metadata to this dataset you will have to import the **.fam** file into **StructuRly**, which generally accompanies **.bed** and **.ped** files. **StructuRly** will automatically use the first two variables of the **.fam** file, which are generally used to ensure the sample identifier and the user-defined population respectively.
#### Section 3: Compare partitions
The third section uses the first two sections input files to compare the partitions obtained from the hierarchical and Bayesian cluster analysis. Obviously, the imported datasets must refer to data of same nature and the number of observations must be the same in both files. The samples cluster memberships of the admixture ancestry analysis partition are assigned considering the highest value of ancestry found in a specific population (*STRUCTURE* or *ADMIXTURE* cluster) for each sample. It means that this partition will divide the observations in the same number of clusters chosen for the population analysis, but if the admixture ancestry is the lowest for a particular subpopulation, this cluster will not be shown in the comparison plot and table, because there are no observations assigned to it.
### Outputs download
The following image shows the main output downloaded from **StructuRly**, the barplot of the ancestry admixture. The sample labels on the X axis are colored according to the population indicated in the user input file, while the symbols at the top of the plot indicate the sampling site. In **StructuRly** there are 25 different symbols available but you can also simply decide to split the entire plot on the basis of the different categories inside the **Pop_ID** and **Loc_ID** variables.
All **StructuRly** outputs can be downloaded as images in various high-quality formats directly from the user interface. However, to download the graphs related to the Triangle plot, obtained through a specific function of the *plotly* package (and not with those of *ggplot2*) you need to download the **orca** software in your computer and follow the instructions at this [link](https://github.com/plotly/orca#installation). If you don't install the **orca** software you can always download the Triangle plot through the functionalities of the *plotly* package through the commands displayed directly on the interactive plot.
**N. B.**: for a dataset with a high sample number (> 500) remember to re-size your plot (width, height and resolution) to better distinguish the bars and the relative IDs.
### Example
Here's a link to the [YouTube video of StructuRly](https://youtu.be/0FUFM6GNaYI) showing an example of using of the software. Moreover, the flowchart below, accessible from the **Instructions** panel of the application, schematize a tutorial to use the software.
### Known bugs and limitations
- in the interactive barplot, the X-axis labels are not colored according to the different populations entered by the user in the input file. To view colored labels, download the image in one of the different formats available. This bug is related to the functions of a third-party package and has been reported to the GitHub community at this [link](https://github.com/ropensci/plotly/issues/1328);
- for more than 40 predefined populations present in the dataset to produce the barplot, the colors used to distinguish such populations within the labels of the X-axis of the barplot could start to repeat;
- there is a limit of 50 Mb for data uploading;
- in the comparison plot, the separation lines of the heatmap cells are not visible when using the interactive graph for a bug present in the package used to produce this output. Again, download the output to view the complete chart;
- when using a large number of populations (> 60) or collection sites defined a priori by the user, the graphics output produced when the observations of the barplot are divided into sections according to the different populations or collection sites (or both) may not be accurate. Moreover, in this case, you could see a slight overlap between the axis title and the axis text of the barplot;
- when using the online version of StructuRly, in order to download the code to produce a plot click on "Download R Code" and then select the code in the dashboard, then copy it (Cmd + C on macOS, CTRL + C on Windows and Ubuntu). Then button to directly copy some content to the user clipboard in an interactive session is still in development.
The slight bugs related to some characteristics of the graphs are shown only inside the interactive plots, but the downloaded file won't present any problem.
### Citation
**StructuRly** was firstly presented during the [International BBCC meetings](https://www.bbcc-meetings.it) held in Naples (Italy) in November 2018 and its implementation has been described inside the paper *StructuRly: a novel shiny app to produce comprehensive, detailed and interactive plots for population genetic analysis* (submitted). If you use this package for your research please cite:
- Criscuolo, N. G. & Angelini, C. StructuRly: A novel shiny app to produce comprehensive, detailed and interactive plots for population genetic analysis. PLoS One 15, e0229330 (2020). [https://doi.org/10.1371/journal.pone.0229330](https://doi.org/10.1371/journal.pone.0229330).
### Contact
For additional information about **StructuRly**, please consult the documentation or [email us](mailto:nico.criscuolo1618@gmail.com).