Title: Deep Learning Prediction of Effectors
Version: 0.1.1
Description: A tool that contains trained deep learning models for predicting effector proteins. 'deepredeff' has been trained to identify effector proteins using a set of known experimentally validated effectors from either bacteria, fungi, or oomycetes. Documentation is available via several vignettes, and the paper by Kristianingsih and MacLean (2020) <doi:10.1101/2020.07.08.193250>.
License: MIT + file LICENSE
URL: https://github.com/ruthkr/deepredeff/
BugReports: https://github.com/ruthkr/deepredeff/issues/
Depends: R (≥ 2.10)
Imports: Biostrings, dplyr, ggplot2, ggthemes, keras, magrittr, purrr, reticulate, rlang, seqinr, tensorflow
Suggests: covr, kableExtra, knitr, rmarkdown, stringr, testthat
VignetteBuilder: knitr
Config/reticulate: list( packages = list( list(package = "tensorflow", version = "2.0.0", pip = FALSE) ) )
Encoding: UTF-8
RoxygenNote: 7.1.1
NeedsCompilation: no
Packaged: 2021-07-16 09:12:16 UTC; kristiar
Author: Ruth Kristianingsih ORCID iD [aut, cre, cph]
Maintainer: Ruth Kristianingsih <ruth.kristianingsih30@gmail.com>
Repository: CRAN
Date/Publication: 2021-07-16 09:30:02 UTC

Pipe operator

Description

See magrittr::%>% for details.

Usage

lhs %>% rhs

Convert AAString class to dataframe

Description

Convert AAString class to dataframe

Usage

aas_to_df(aas)

Arguments

aas

AAString class object.

Value

Returns data in data frame.

Examples

input_fasta <- system.file("extdata/example/fungi_sample.fasta", package = "deepredeff")
input_aas <- Biostrings::readAAStringSet(input_fasta)
input_df <- aasset_to_df(input_aas)

Convert AAStringset class to dataframe

Description

Convert AAStringset class to dataframe

Usage

aasset_to_df(aas)

Arguments

aas

AAStringset class object.

Value

Returns data in data frame.

Examples

input <- "MSHMTFNTWKAGLWRLAAAAVLSLLPVVARAAVPGITGPTFDLTAQPGRANQPDGASVYSWGYGCNPRTVPGFLPSVNPLAGQ"
input_aas <- Biostrings::AAString(input)
input_df <- aas_to_df(input_aas)

deepredeff package

Description

Effector protein predictor using Deep Learning models.

Details

See the README on GitHub

Integer encoder

Description

encode_integer integer-encodes sequence in a string format.

Usage

encode_integer(sequence, max_length = 4034)

Arguments

sequence

Sequence in a string format.

max_length

Maximum length of sequence to encode.

Value

Integer encoded sequence.

Examples

sample_seq <- "MSHMTFNTWKAGLWRLAAAAVLSLLPVVARAAVPGITGPTFDLTAQPGRANQPDGASVYSWGYGCNPRTVPGFLPSVNPLAGQ"
encoded_seq <- encode_integer(sample_seq)

One-hot encoder

Description

encode_one_hot one-hot-encodes sequence in a string format.

Usage

encode_one_hot(sequence, max_length = 4034)

Arguments

sequence

Sequence in a string format.

max_length

Maximum length of sequence to encode.

Value

One-hot encoded sequence.

Examples

sample_seq <- "MSHMTFNTWKAGLWRLAAAAVLSLLPVVARAAVPGITGPTFDLTAQPGRANQPDGASVYSWGYGCNPRTVPGFLPSVNPLAGQ"
encoded_seq <- encode_one_hot(sample_seq)

Weighted ensemble

Description

Weighted ensemble

Usage

ensemble_weighted(pred_list, weights)

Arguments

pred_list

List of deep learning models.

weights

Accuracy values from evaluation on the validation dataset.

Value

Returns the prediction results from weighted ensemble.

Convert fasta format to dataframe

Description

Convert fasta format to dataframe

Usage

fasta_to_df(fasta_path)

Arguments

fasta_path

Path of FASTA file.

Value

Returns data in data frame.

Examples

input <- system.file("extdata/example/fungi_sample.fasta", package = "deepredeff")
input_df <- fasta_to_df(input)

Get ensemble methods

Description

This function is used to get the ensemble methods used for each taxon group. If weights are needed for a particular ensemble, then the weights will automatically follow.

Usage

get_ensemble_method(taxon)

Arguments

taxon

taxon group

Value

Returns ensemble method and weights.

Install the TensorFlow backend

Description

TensorFlow will be installed into an "r-tensorflow" virtual or conda environment. Note that "virtualenv" is not available on Windows (as this isn't supported by TensorFlow).

Usage

install_tensorflow(
  method = c("conda", "virtualenv"),
  conda = "auto",
  version = "2.0.0",
  extra_packages = NULL,
  ...
)

Arguments

method

Installation method ("conda" or "virtualenv").

conda

The path to a conda executable. Use "auto" to allow reticulate to automatically find an appropriate conda binary. See Finding Conda for more details.

version

TensorFlow version to install ( by default, "2.0.0").

extra_packages

Additional PyPI packages to install along with TensorFlow.

...

Other arguments passed to reticulate::py_install().

Custom Installation

Custom installations of TensorFlow are not supported yet by deepredeff.

Additional Packages

If you wish to add additional PyPI packages to your TensorFlow environment you can either specify the packages in the extra_packages argument of install_tensorflow(), or alternatively install them into an existing environment using the reticulate::py_install() function.

Notice that this may have undesired side-effects on Windows installations.

Examples

## Not run: 
# Default installation
library(deepredeff)
install_tensorflow()

## End(Not run)

Load model

Description

load_model() loads model saved in hdf5 format

Usage

load_model(taxon = c("bacteria", "fungi", "oomycete"))

Arguments

taxon

Name of taxon.

Value

Returns all of the hyperparamaters and parameters of particular model from specific taxon chosen.

Wildcard Expansion on File Paths

Description

Wildcard Expansion on File Paths

Usage

package_glob(..., pattern)

Arguments

...

Path

pattern

Pattern

Value

Glob

Plot the results of prediction

Description

Plot the results of prediction

Usage

## S3 method for class 'tbl_deepredeff'
plot(x, ...)

Arguments

x

tbl_deepredeff object

...

additional arguments ignored.

Value

class distribution plot

Examples


# FASTA input
input_fasta <- system.file("extdata/example/fungi_sample.fasta", package = "deepredeff")

pred_result <- deepredeff::predict_effector(
  input = input_fasta,
  taxon = "fungi"
)

plot(pred_result)

Predict effector

Description

predict_effector is used to predict effector protein given amino acid protein sequences.

Usage

predict_effector(input, taxon)

## S3 method for class 'character'
predict_effector(input, taxon)

## S3 method for class 'data.frame'
predict_effector(input, taxon)

## S3 method for class 'AAStringSet'
predict_effector(input, taxon)

## S3 method for class 'AAString'
predict_effector(input, taxon)

## Default S3 method:
predict_effector(input, taxon)

Arguments

input

Input data that contains amino acid sequence(s). It can be in fasta format, strings, AAString, AAStringset, and dataframe.

taxon

Taxon group of input data. Available taxons are bacteria, fungi, and oomycete.

Value

predict_effector returns an object of class "tbl_deepredeff" or for multiple responses of class c("tbl_deepredeff", "data.frame").

An object of class "tbl_deepredeff" is a data frame containing at least the following components:

sequence

the sequence(s) from the input data.

s_score

score obtained from sigmoid function showing how likely the sequences to be an effector.

prediction

class prediction for each sequence, obtained from s_score. If the value of s_score >= 0.5, it will be classified as an effector. Otherwise, it will be classified as a non-effector.

Examples


# FASTA input
input_fasta <- system.file("extdata/example/fungi_sample.fasta", package = "deepredeff")

pred_result <- deepredeff::predict_effector(
  input = input_fasta,
  taxon = "fungi"
)

Prediction mapper helper

Description

Prediction mapper helper

Usage

prediction_mapper(sequence_list, model_list)

Arguments

sequence_list

List of sequences input.

model_list

List of models.

Value

Returns list of prediction result of each sequence.

Get the summary of the prediction results

Description

Get the summary of the prediction results

Usage

## S3 method for class 'tbl_deepredeff'
summary(object, ...)

Arguments

object

Results of prediction from deepredeff::predict_effector().

...

Additional arguments ignored.

Examples


# FASTA input
input_fasta <- system.file("extdata/example/fungi_sample.fasta", package = "deepredeff")

pred_result <- deepredeff::predict_effector(
  input = input_fasta,
  taxon = "fungi"
)

summary(pred_result)