diff_analysis: Differential expression analysis

View source: R/method-diff-analysis.R

diff_analysisR Documentation

Differential expression analysis

Description

Differential expression analysis

Usage

diff_analysis(obj, ...)

## S3 method for class 'data.frame'
diff_analysis(
  obj,
  sampleda,
  classgroup,
  subclass = NULL,
  taxda = NULL,
  alltax = TRUE,
  include.rownames = FALSE,
  standard_method = NULL,
  mlfun = "lda",
  ratio = 0.7,
  firstcomfun = "kruskal.test",
  padjust = "fdr",
  filtermod = "pvalue",
  firstalpha = 0.05,
  strictmod = TRUE,
  fcfun = "generalizedFC",
  secondcomfun = "wilcox.test",
  clmin = 5,
  clwilc = TRUE,
  secondalpha = 0.05,
  subclmin = 3,
  subclwilc = TRUE,
  ldascore = 2,
  normalization = 1e+06,
  bootnums = 30,
  ci = 0.95,
  type = "species",
  ...
)

## S3 method for class 'phyloseq'
diff_analysis(obj, ...)

Arguments

obj

object,a phyloseq class contained otu_table, sample_data, taxda, or data.frame, nrow sample * ncol features.

...

additional parameters.

sampleda

data.frame, nrow sample * ncol factor, the sample names of sampleda and data should be the same.

classgroup

character, the factor name in sampleda.

subclass

character, the factor name in sampleda, default is NULL, meaning no subclass compare.

taxda

data.frame, the classification of the feature in data. default is NULL.

alltax

logical, whether to set all classification (taxonomy) as features when taxda is not NULL, default is TRUE.

include.rownames

logical, whether to consider the OTU of obj as (all taxonomy) features, when taxda is not NULL, default is FALSE.

standard_method

character, the method of standardization, see also decostand, default is NULL, it represents that the relative abundance of taxonomy will be used. If count was set, it represents the count reads of taxonomy will be used.

mlfun

character, the method for calculating the effect size of features, choose "lda" or "rf", default is "lda".

ratio

numeric, range from 0 to 1, the proportion of samples for calculating the effect size of features, default is 0.7.

firstcomfun

character, the method for first test, "oneway.test" for normal distributions, suggested choosing "kruskal.test" for uneven distributions, default is "kruskal.test", or you can use lm, glm, or glm.nb (for negative binomial distribution), or 'kruskal_test', 'oneway_test' of 'coin'.

padjust

character, the correction method, default is "fdr".

filtermod

character, the method to filter, default is "pvalue".

firstalpha

numeric, the alpha value for the first test, default is 0.05.

strictmod

logical, whether to performed in one-against-one, default is TRUE (strict).

fcfun

character, default is "generalizedFC", it can't be set another at the present time.

secondcomfun

character, the method for one-against-one, default is "wilcox.test" for uneven distributions, or 'wilcox_test' of 'coin', or you can also use 'lm', 'glm', 'glm.nb'(for negative binomial distribution in 'MASS').

clmin

integer, the minimum number of samples per classgroup for performing test, default is 5.

clwilc

logical, whether to perform test of per classgroup, default is TRUE.

secondalpha

numeric, the alpha value for the second test, default is 0.05.

subclmin

integer, the minimum number of samples per suclass for performing test, default is 3.

subclwilc

logical, whether to perform test of per subclass, default is TRUE, meaning more strict.

ldascore

numeric, the threshold on the absolute value of the logarithmic LDA score, default is 2.

normalization

integer, set the normalization value, set a big number if to get more meaningful values for the LDA score, or you can set NULL for no normalization, default is 1000000.

bootnums

integer, set the number of bootstrap iteration for lda or rf, default is 30.

ci

numeric, the confidence interval of effect size (LDA or MDA), default is 0.95.

type

character, the type of datasets, default is "species", if the dataset is not about species, such as dataset of kegg function, you should set it to "others".

Value

diff_analysis class.

Author(s)

Shuangbin Xu

Examples

## Not run: 
data(kostic2012crc)
kostic2012crc %<>% as.phyloseq()
head(phyloseq::sample_data(kostic2012crc),3)
kostic2012crc <- phyloseq::rarefy_even_depth(kostic2012crc,rngseed=1024)
table(phyloseq::sample_data(kostic2012crc)$DIAGNOSIS)
set.seed(1024)
diffres <- diff_analysis(kostic2012crc, classgroup="DIAGNOSIS",
                        mlfun="lda", filtermod="fdr",
                        firstcomfun = "kruskal.test",
                        firstalpha=0.05, strictmod=TRUE,
                        secondcomfun = "wilcox.test",
                        subclmin=3, subclwilc=TRUE,
                        secondalpha=0.01, ldascore=3)

## End(Not run)

YuLab-SMU/MicrobiotaProcess documentation built on July 26, 2024, 4:21 a.m.