Recipe View

Perform eDNA Occupancy Modeling

This recipe runs eDNAoccupancy R package to estimate the occurrence of a species by modeling eDNA d…

2 results • updated 4.4 years ago by Istvan Albert

This recipe runs eDNAoccupancy R package to estimate the occurrence of a species by modeling eDNA data.

eDNA occupancy: An R package for multi‐scale occupancy modeling of environmental DNA data by Dorazio, R. M., & Erickson, R. A., Molecular Ecology Resources, 18, 368–380 (2017).

Required inputs

Presence/absence eDNA detection data in CSV format.
Metadata file in CSV format.

Detection file format

In the presence/absence table, the site column should have the header label site, the column specifying sample should have the sample sample. The rest of the headers should be sample names. The presence and absence values should be encoded with 0 and 1.

An example presence/absence table:

site,sample,pcr1,pcr2,pcr3,pcr4,pcr5,pcr6
Hunter_Creek,1,0,0,0,0,0,0
Pistol_River,1,0,0,0,0,0,0
Winchuck_River,1,0,0,0,0,0,0
Tillas_Slough,1,0,0,0,0,0,0
Lake_Earl,1,1,1,1,1,1,1
Elk_Creek,1,0,0,0,0,0,0

Metadata format:

The metadata headers should list the various labels associated with the site. The labels will be used as parameter entries in the interface.

site,twg,sal,turb,fish,veg
Hunter_Creek,0,7.2,60,0,0
Pistol_River,0,13,50,0,0
Winchuck_River,0,2,43,0,0
Tillas_Slough,17.8,9,236.8,160.9,1
Lake_Earl,1.06,4,124.15,113,1
Elk_Creek,0,30,144.67,19.3,0

In this example the usable paramters then would be twg, sal, turb, fish and veg.

Parameter selection tips

While running the recipe, ~ should prefix the covariate specification.
Enter ~1 in the text field if a covariate is none.
Multiple covariates can be combined using +.
An example covariate specification is ~sal+fish.

Covariates entered should match the column headers in the metadata file.

Copy recipe

You need write access to the original recipe to edit.

# Load R package
library(eDNAoccupancy)

# Input variables
eDNA_detection_file="Goby-Data.csv"
metadata_file="Goby-Metadata.csv"
site_covariates = "~ veg"
sample_covariates = "~ sal+fish+turb"
site_sample_covariates ="~ sal+twg"

# Output files
posterior_summary_file = "posterior_summary.txt"
posterior_median_estimates_file = "posterior_median_estimates.txt"
pdf_file ="plot.pdf"

# Function to plot the relationship between a covariate and the probability of eDNA detection in samples.
plot_covariate= function(metadata,covariate)
{
  vals = metadata[, covariate]
  plot(vals, p$median[,1], ylim=ylimits, cex=3, main=covariate, xlab=covariate, cex.lab=1.5)
  segments(vals, p$lower[,1], vals, p$upper[,1], lwd=2)
}

# Read input data and meta data.
detection_data = read.table(eDNA_detection_file, header=TRUE, sep="," )
metadata = read.table(metadata_file, header=TRUE, sep="," )

# Compute occupancy data matrices
species_detections = occData(detection_data, siteColName = 'site',
                             sampleColName = 'sample')

# Center and scale numeric-valued covariate measurements
metadata.sc = scaleData(metadata)

# Fit the model.
fit = occModel(formulaSite = site_covariates, formulaSiteAndSample = site_sample_covariates, formulaReplicate = sample_covariates ,
               detectionMats = species_detections, siteData = metadata.sc, siteColName = 'site')

# Print Posterior Summaries
posterior.summary=as.data.frame(posteriorSummary(fit, burnin=1000, mcError=TRUE,outputSummary = TRUE))
posterior.summary.df = data.frame("params"=rownames(posterior.summary),posterior.summary)

# Write posterior summaries to file.
write.table(posterior.summary.df,file = posterior_summary_file, sep="\t", row.names=FALSE, quote=FALSE)

# Estimate the posterior summaries of derived parameters of the models - psi, theta and p
# psi = Probability of occurrence of a species in a location 
# theta = Conditional probability of occurrence of a species in a sample given that species is present in the location where the sample is collected. 
# p = Conditional probability of species detection in a subsample of a sample given that species is present in the sample. 

psi = posteriorSummaryOfSiteOccupancy(fit, burnin=1000)
theta = posteriorSummaryOfSampleOccupancy(fit, burnin=1000)
p = posteriorSummaryOfDetection(fit, burnin=1000)

# output estimates of posterior medians
post.medians=cbind(psi=psi$median, theta=theta$median[,1], p=p$median[,1])

# Write estimates of posterior medians to a file.
post.medians.df = data.frame("site"=rownames(post.medians), post.medians)
write.table(post.medians.df,file = posterior_median_estimates_file, sep="\t", row.names=FALSE, quote=FALSE)

# Extract numeric ovariates to plot
covariates_to_plot =metadata[sapply(metadata,is.double)]
col_names = colnames(covariates_to_plot)
ncols= ncol(covariates_to_plot)

# Plot the estimated relationship between each covariate (eg: sal, fish, or turb) 
# and the probability of detection of eDNA in PCR replicates 

pdf(pdf_file, width=10, height=10)
ylimits = c(0.2,1)
par(mfrow=c(1,ncols), mar = c(5, 5, 5, 5), oma=c(0,0,3,0))

for (col in col_names) {
  plot_covariate(metadata,col)
}

mtext("Estimated relationship between covariate and the probability of eDNA detection", outer=TRUE, cex=1.5)
dev.off()
unlink("mc.*")

You need write access to the original recipe to edit.

Click the buttons on the right to create new fields.

Add text field Add float field Add data field Add checkbox Add dropdown Add upload field Add integer field Add radio button

Edit the content of each interface element.

[data]
value = "Goby-Data.csv"
label = "eDNA detection data"
help = "Presence/Absence table of eDNA detection in samples in each location."
source = "PROJECT"
display = "DROPDOWN"
type = "eDNA"

[metadata]
value = "Goby-Metadata.csv"
label = "Metadata"
source = "PROJECT"
display = "DROPDOWN"
help = "Table with covariates that are informative of eDNA detection in each location."
type = "METADATA"

[site_covariates]
label = "Site Covariates"
display = "TEXTBOX"
help = "List of site covariates separated by \"+\". This is used to specify a model for the probability of occurrence of eDNA at a location. Enter ~1 if none."
value = "~ veg"
regex = ".+"

[sample_covariates]
label = "Sample Covariates"
display = "TEXTBOX"
help = "List of sample covariates separated by \"+\". This is used to specify a model model of detection in replicates of a sample. Enter ~1 if none."
value = "~ sal+fish+turb"
regex = ".+"

[site_sample_covariates]
label = "Site and Sample Covariates"
display = "TEXTBOX"
help = "List of covariates separated by \"+\". This is used to specify a model of occurrence in different samples of a site. Enter ~1 if none."
value = "~ sal+twg"
regex = ".+"

[settings.execute]
command = "Rscript recipe.sh"

You need write access to the original recipe to edit.

Name

Recipe display name

Identifier

Unique identifier for the recipe.

Image :

Optional image for the recipe ( 500px Maximum ).

Rank:

Used to order recipes (optional).

This recipe runs eDNAoccupancy R package to estimate the occurrence of a species by modeling eDNA data.

* [eDNA occupancy: An R package for multi‐scale occupancy modeling of environmental DNA data][1] by Dorazio, R. M., & Erickson, R. A., Molecular Ecology Resources, 18, 368–380 (2017).

[1]: https://onlinelibrary.wiley.com/doi/full/10.1111/1755-0998.12735

### Required inputs

1. Presence/absence eDNA detection data in CSV format.
2. Metadata file in CSV format.

### Detection file format

In the presence/absence table, the site column should have the header label `site`, the column specifying sample should have the sample `sample`. The rest of the headers should be sample names. The presence and absence values should be encoded with `0` and `1`.

An example presence/absence table:

```
site,sample,pcr1,pcr2,pcr3,pcr4,pcr5,pcr6
Hunter_Creek,1,0,0,0,0,0,0
Pistol_River,1,0,0,0,0,0,0
Winchuck_River,1,0,0,0,0,0,0
Tillas_Slough,1,0,0,0,0,0,0
Lake_Earl,1,1,1,1,1,1,1
Elk_Creek,1,0,0,0,0,0,0

```

### Metadata format:

The metadata headers should list the various labels associated with the site. The labels will be used as parameter entries in the interface.

```
site,twg,sal,turb,fish,veg
Hunter_Creek,0,7.2,60,0,0
Pistol_River,0,13,50,0,0
Winchuck_River,0,2,43,0,0
Tillas_Slough,17.8,9,236.8,160.9,1
Lake_Earl,1.06,4,124.15,113,1
Elk_Creek,0,30,144.67,19.3,0
```

In this example the usable paramters then would be `twg`, `sal`, `turb`, `fish` and `veg`.

### Parameter selection tips

- While running the recipe, `~` should prefix the covariate specification. 
- Enter `~1` in the text field if a covariate is none. 
- Multiple covariates can be combined using `+`. 
- An example covariate specification is   `~sal+fish`.

Covariates entered should match the column headers in the metadata file.

Insert Image

From the web

From your computer

Cancel

Back