Title: A Machine-Learning Based Tool to Automate the Identification of Biological Database IDs
Version: 1.0.4
Description: The number of biological databases is growing rapidly, but different databases use different IDs to refer to the same biological entity. The inconsistency in IDs impedes the integration of various types of biological data. To resolve the problem, we developed 'MantaID', a data-driven, machine-learning based approach that automates identifying IDs on a large scale. The 'MantaID' model's prediction accuracy was proven to be 99%, and it correctly and effectively predicted 100,000 ID entries within two minutes. 'MantaID' supports the discovery and exploitation of ID patterns from large quantities of databases. (e.g., up to 542 biological databases). An easy-to-use freely available open-source software R package, a user-friendly web application, and API were also developed for 'MantaID' to improve applicability. To our knowledge, 'MantaID' is the first tool that enables an automatic, quick, accurate, and comprehensive identification of large quantities of IDs, and can therefore be used as a starting point to facilitate the complex assimilation and aggregation of biological data across diverse databases.
License: GPL (≥ 3)
Encoding: UTF-8
RoxygenNote: 7.3.2
Depends: R (≥ 4.4.0),biomaRt,caret,keras,mlr3tuning,mlr3
LazyData: true
URL: https://molaison.github.io/MantaID/
Imports: ggplot2,data.table,magrittr,stringr,tibble,tidyr,tidyselect,ggcorrplot,reshape2,scutr,paradox,RColorBrewer,purrr,dplyr
Suggests: mlr3hyperband,mlr3learners,ranger,rpart,xgboost
NeedsCompilation: no
Packaged: 2024-09-09 07:07:26 UTC; pc
Author: Zhengpeng Zeng ORCID iD [aut, cre, ctb], Longfei Mao ORCID iD [aut, cph], Feng Yu ORCID iD [aut], Jiamin Hu ORCID iD [ctb], Xiting Wang ORCID iD [ctb]
Maintainer: Zhengpeng Zeng <molaison@foxmail.com>
Repository: CRAN
Date/Publication: 2024-09-09 07:50:02 UTC

ID example dataset.

Description

ID example dataset.

Usage

Example

Format

A tibble with 5000 rows and 2 variables.

ID

A identifier character.

class

The database the ID belongs to.

A wrapper function that executes MantaID workflow.

Description

A wrapper function that executes MantaID workflow.

Usage

mi(
  cores = NULL,
  levels = c("*", 0:9, letters, LETTERS, "_", ".", "-", " ", "/", "\\", ":"),
  ratio = 0.3,
  para_blc = FALSE,
  model_path = NULL,
  batch_size = 128,
  epochs = 64,
  validation_split = 0.3,
  graph_path = NULL
)

Arguments

cores

The number of cores used when balancing data.

levels

The vector that includes all the single characters occurred in IDs.

ratio

The ratio of the test set.

para_blc

A logical value whether using parallel computing when balancing data.

model_path

The path to save models.

batch_size

The batch size of deep learning model fitting.

epochs

The epochs of deep learning model fitting.

validation_split

The validation ratio of deep learning model fitting.

graph_path

The path to save graphs.

Value

The list of models and graphs.

Data balance. Most classes adopt random undersampling, while a few classes adopt smote method to oversample to obtain relatively balanced data.

Description

Data balance. Most classes adopt random undersampling, while a few classes adopt smote method to oversample to obtain relatively balanced data.

Usage

mi_balance_data(data, ratio = 0.3, parallel = FALSE)

Arguments

data

A data frame. Except for class column, all are numeric types.

ratio

Numeric between 0 and 1. The percent of the test set split from data.

parallel

Logical.

Value

A list contains a train set and a test set.

Reshape data and delete meaningless rows.

Description

Reshape data and delete meaningless rows.

Usage

mi_clean_data(data, cols = everything(), placeholder = c("-"))

Arguments

data

A dataframe or tibble or data.table or matrix. Names of the column will be regard as the class of ID included in column.

cols

Character vectors. Columns of data that contain the IDs

placeholder

Character vectors. IDs included in placeholder will be omitted.

Value

A tibble with two columns("ID" and "class")

Examples

data <- tibble::tibble(
  "class1" = c("A", "B", "C", "D"),
  "class2" = c("E", "F", "G", "H"),
  "class3" = c("L", "M", "-", "O")
)
mi_clean_data(data)

ID-related datasets in biomart.

Description

ID-related datasets in biomart.

Usage

mi_data_attributes

Format

A dataframe with 65 variables and 3 variables.

name

The name of dataset.

description

Description of dataset.

page

collection of attributes.

Processed ID data.

Description

Processed ID data.

Usage

mi_data_procID

Format

A tibble dataframe with 5000 rows and 21 variables.

pos1 to pos20

Splited ID.

class

The databases that ID belongs to.

ID dataset for testing.

Description

ID dataset for testing.

Usage

mi_data_rawID

Format

A tibble with 5000 rows and 2 variables.

ID

A identifier character.

class

The database the ID belongs to.

Performing feature selection in a automatic way based on correlation and feature importance.

Description

Performing feature selection in a automatic way based on correlation and feature importance.

Usage

mi_filter_feat(data, cor_thresh = 0.7, imp_thresh = 0.99, union = FALSE)

Arguments

data

The data frame returned by mi_to_numer().

cor_thresh

The threshold set for Pearson correlation. If correlation value is over this threshold, the two features will be viewed as redundant and one of them will be removed.

imp_thresh

The threshold set for feature importance. The last several features with the lowest importance will be removed if remained importance lower than imp_thresh.

union

The method for combining the decisions of correlation method and importance method. If TRUE, any of the features calculated by the two methods will be returned. Otherwise, only features in the results of both methods will be returned.

Value

The names of the features that should be removed.

Get ID data from the Biomart database using attributes.

Description

Get ID data from the Biomart database using attributes.

Usage

mi_get_ID(
  attributes,
  biomart = "genes",
  dataset = "hsapiens_gene_ensembl",
  mirror = "asia"
)

Arguments

attributes

A dataframe.The information we want to retrieve.Use mi_get_ID_attr to hava try.

biomart

BioMart database name you want to connect to. Use biomaRt::listEnsembl to retrieve the possible database names.

dataset

Datasets of the selected BioMart database.

mirror

Specify an Ensembl mirror to connect to.

Value

A tibble dataframe.

Get ID attributes from the Biomart database.

Description

Get ID attributes from the Biomart database.

Usage

mi_get_ID_attr(
  biomart = "genes",
  dataset = "hsapiens_gene_ensembl",
  mirror = "asia"
)

Arguments

biomart

BioMart database name you want to connect to.Use biomaRt::listEnsembl to retrieve the possible database names.

dataset

Datasets of the selected BioMart database.

mirror

Specify an Ensembl mirror to connect to.

Value

A dataframe.

Compute the confusion matrix for the predicted result.

Description

Compute the confusion matrix for the predicted result.

Usage

mi_get_confusion(result_list, ifnet = FALSE)

Arguments

result_list

A list returned from model training functions.

ifnet

Logical.Whether the data is obtained by a deep learning model.

Value

A confusionMatrix object.

Plot the bar plot for feature importance.

Description

Plot the bar plot for feature importance.

Usage

mi_get_importance(data)

Arguments

data

A table.

Value

A bar plot.

Observe the distribution of the false response of the test set.

Description

Observe the distribution of the false response of the test set.

Usage

mi_get_miss(predict)

Arguments

predict

An R6 class PredictionClassif.

Value

A tibble data frame that records the number of wrong predictions for each category ID

Get max length of ID data.

Description

Get max length of ID data.

Usage

mi_get_padlen(data)

Arguments

data

A dataframe.

Value

An int object.

Examples

data(mi_data_rawID)
mi_get_padlen(mi_data_rawID)

Plot correlation heatmap.

Description

Plot correlation heatmap.

Usage

mi_plot_cor(data, cls = "class")

Arguments

data

Data frame including IDs' position features.

cls

The name of the class column.

Value

A heatmap.

Examples

data(mi_data_procID)
data_num <- mi_to_numer(mi_data_procID)
mi_plot_cor(data_num)

Plot heatmap for result confusion matrix.

Description

Plot heatmap for result confusion matrix.

Usage

mi_plot_heatmap(table, name = NULL, filepath = NULL)

Arguments

table

A table.

name

Model names.

filepath

File path the plot to save. Default NULL.

Value

A ggplot object.

Predict new data with a trained learner.

Description

Predict new data with a trained learner.

Usage

mi_predict_new(data, result, ifnet = F)

Arguments

data

A dataframe.

result

The result object from a previous training.

ifnet

A boolean indicating if a neural network is used for prediction.

Value

A data frame that contains features and 'predict' class.

Compare classification models with small samples.

Description

Compare classification models with small samples.

Usage

mi_run_bmr(data, row_num = 1000, resamplings = rsmps("cv", folds = 10))

Arguments

data

A tibble.All are numeric except the first column is a factor.

row_num

The number of samples used.

resamplings

R6/Resampling.Resampling method.

Value

A list of R6 class of benchmark results and scores of test set. examples data(mi_data_procID) mi_run_bmr(mi_data_procID)

Cut the string of ID column character by character and divide it into multiple columns.

Description

Cut the string of ID column character by character and divide it into multiple columns.

Usage

mi_split_col(data, cores = NULL, pad_len = 10)

Arguments

data

Dataframe(tibble) to be split.

cores

Int.The num of cores to allocate for computing.

pad_len

The length of longest id, i.e. the maxlength.

Value

A tibble with pad_len+1 column.

Split the string into individual characters and complete the character vector to the maximum length.

Description

Split the string into individual characters and complete the character vector to the maximum length.

Usage

mi_split_str(str, pad_len)

Arguments

str

The string to be splited.

pad_len

The length of longest ID, i.e. the maxlength.

Value

Splited character vector.

Examples

string_test <- "Good Job"
length <- 15
mi_split_str(string_test, length)

Convert data to numeric, and for the ID column convert with fixed levels.

Description

Convert data to numeric, and for the ID column convert with fixed levels.

Usage

mi_to_numer(
  data,
  levels = c("*", 0:9, letters, LETTERS, "_", ".", "-", " ", "/", "\\", ":")
)

Arguments

data

A tibble with n position column(pos1,pos2,...) and class column.

levels

Characters accommodated in IDs.

Value

A numeric data frame with numerical or factor type columns.

Examples

data(mi_data_procID)
mi_to_numer(mi_data_procID)

Train a three layers neural network model.

Description

Train a three layers neural network model.

Usage

mi_train_BP(
  train,
  test,
  cls = "class",
  path2save = NULL,
  batch_size = 128,
  epochs = 64,
  validation_split = 0.3,
  verbose = 0
)

Arguments

train

A dataframe with the class column as label.

test

A dataframe with the class column as label.

cls

A character.The name of the label column.

path2save

The folder path to store the model and train history.

batch_size

Integer or NULL. The number of samples per gradient update.

epochs

The number of epochs to train the model.

validation_split

Float between 0 and 1. Fraction of the training data to be used as validation data.

verbose

The verbosity mode.

Value

A list object containing the prediction confusion matrix, the model object, and the mapping of predicted numbers to classes.

Random Forest Model Training.

Description

Random Forest Model Training.

Usage

mi_train_rg(train, test, measure = msr("classif.acc"), instance = NULL)

Arguments

train

A dataframe.

test

A dataframe.

measure

Model evaluation method.

instance

A tuner.

Value

A list of learner for predicting and predicted result of test set.

Classification tree model training.

Description

Classification tree model training.

Usage

mi_train_rp(train, test, measure = msr("classif.acc"), instance = NULL)

Arguments

train

A dataframe.

test

A dataframe.

measure

Model evaluation method.Use mlr_measures and msr() to view and choose metrics.

instance

A tuner.

Value

A list of learner for predicting and predicted result of test set.

Xgboost model training

Description

Xgboost model training

Usage

mi_train_xgb(train, test, measure = msr("classif.acc"), instance = NULL)

Arguments

train

A dataframe.

test

A dataframe.

measure

Model evaluation method.

instance

A tuner.

Value

A list of learner for predicting and predicted result of test set.

Tune the Random Forest model by hyperband.

Description

Tune the Random Forest model by hyperband.

Usage

mi_tune_rg(
  data,
  resampling = rsmp("cv", folds = 5),
  measure = msr("classif.acc"),
  eta = 3
)

Arguments

data

A tibble.All are numeric except the first column is a factor.

resampling

R6/Resampling.

measure

Model evaluation method.Use mlr_measures and msr() to view and choose metrics.

eta

The percent parameter configurations discarded.

Value

A list of tuning instance and stage plot.

Tune the Decision Tree model by hyperband.

Description

Tune the Decision Tree model by hyperband.

Usage

mi_tune_rp(
  data,
  resampling = rsmp("bootstrap", ratio = 0.8, repeats = 5),
  measure = msr("classif.acc"),
  eta = 3
)

Arguments

data

A tibble.All are numeric except the first column is a factor.

resampling

R6/Resampling.

measure

Model evaluation method.Use mlr_measures and msr() to view and choose metrics.

eta

The percent parameter configurations discarded.

Value

A list of tuning instance and stage plot.

Tune the Xgboost model by hyperband.

Description

Tune the Xgboost model by hyperband.

Usage

mi_tune_xgb(
  data,
  resampling = rsmp("cv", folds = 5),
  measure = msr("classif.acc"),
  eta = 3
)

Arguments

data

A tibble.All are numeric except the first column is a factor.

resampling

R6/Resampling.

measure

Model evaluation method.Use mlr_measures and msr() to view and choose metrics.

eta

The percent parameter configurations discarded.

Value

A list of tuning instance and stage plot.

Predict with four models and unify results by the sub-model's specificity score to the four possible classes.

Description

Predict with four models and unify results by the sub-model's specificity score to the four possible classes.

Usage

mi_unify_mod(
  data,
  col_id,
  result_rg,
  result_rp,
  result_xgb,
  result_BP,
  c_value = 0.75,
  pad_len = 30
)

Arguments

data

A dataframe contains the ID column.

col_id

The name of ID column.

result_rg

The result from the Random Forest model.

result_rp

The result from the Decision Tree model.

result_xgb

The result from the XGBoost model.

result_BP

The result from the Backpropagation Neural Network model.

c_value

A numeric value used in the final prediction calculation.

pad_len

The length to pad the ID characters to.

Value

A dataframe.