| Title: | Transcription Factors Association Rules Miner |
|---|---|
| Description: | It searches for relevant associations of transcription factors with a transcription factor target, in specific genomic regions. It also allows to evaluate the Importance Index distribution of transcription factors (and combinations of transcription factors) in association rules. |
| Authors: | Liuba Nausicaa Martino, Alice Parodi, Gaia Ceddia, Piercesare Secchi, Stefano Campaner, Marco Masseroli |
| Maintainer: | Liuba Nausicaa Martino <[email protected]> |
| License: | Artistic-2.0 |
| Version: | 1.29.0 |
| Built: | 2024-10-31 05:44:00 UTC |
| Source: | https://github.com/bioc/TFARM |
DELTA is a list of 12 elements and each element has two columns
representing support (diff_supp_Z) and confidence (diff_conf_Z)
respectively. It is included in the
data_man collection.
data("data_man")data("data_man")
An object of class "list"
# DELTA is found in the data_man collection of datasets: data("data_man") head(DELTA[[1]])# DELTA is found in the data_man collection of datasets: data("data_man") head(DELTA[[1]])
For a set of candidate co-regulator transcription factors, the Importance Index distribution of each transcription factor is plotted in boxplots. The shape of every boxplot depends on: the dimension of the distribution, which is equal to the number of rules in which each transcription factor appears (the higher is such number, the larger is a boxplot), the variability of the distribution (the higher is the variability of the Importance Index distribution, the longer is a boxplot). Moreover, the higher is the median of the Importance Index distribution for a candidate co-regulator transcription factor, the higher the boxplot is aligned with respect to the y-axis.
distribViz(I, TFs)distribViz(I, TFs)
I |
a list of Importance Index distributions (IIDs) |
TFs |
string vector with the names of the transcription factors |
A boxplot representing the transcription factor IIDs
# Load IMP_Z and p from the data_man collection of datasets: data('data_man') # Plot the Importance Index distributions of the transcription factors in p: distribViz(IMP_Z,p_TFs)# Load IMP_Z and p from the data_man collection of datasets: data('data_man') # Plot the Importance Index distributions of the transcription factors in p: distribViz(IMP_Z,p_TFs)
A square matrix (TFs x TFs) is built. The element (i,j) of such matrix contains the mean Importance Index (II) of the couple of transcription factors (TFi, TFj). This is represented in a heatmap visualization, where the scale color from blue to white indicates low mean importance of the couple and from white to red indicates high mean importance of the couple.
heatI(TFs, I)heatI(TFs, I)
TFs |
a string vector with names of transcription factors. |
I |
a vector of mean II of pairs of transcription factors in TFs. |
The Importance index of pairs of Transcription Factors as a heatmap
# Load p_TFs and I_c_2 from the data_man collection of datasets: data('data_man') # Heatmap visualization of the mean importances of transcription factors in p # and their combinations in two elements: heatI(p_TFs, I_c_2)# Load p_TFs and I_c_2 from the data_man collection of datasets: data('data_man') # Heatmap visualization of the mean importances of transcription factors in p # and their combinations in two elements: heatI(p_TFs, I_c_2)
Within the data_man data collection, the dataset I_c_2 has 2 columns
and 78 rows:
the fist column (TF) contains the transcription factors
(single or in pairs) and the second column (imp) lists the Importance
Indexes associated with each trascription factor or pair of transcription
factors.
data("data_man")data("data_man")
An object of class "data.frame"
# I_c_2 is found in the data_man collection of datasets: data("data_man") head(I_c_2$imp)# I_c_2 is found in the data_man collection of datasets: data("data_man") head(I_c_2$imp)
Given an association rule and a transcription factor TFi,
it is evaluated the contribution of TFi in the rule for the prediction
of the presence of the item in the right-hand-side of the rule.
Since this contribution is evaluated based on the variations of
support and confidence of the rule, the user can visualize
such variations by setting the parameter figures = TRUE.
IComp(TFi, rules_TF, rules_noTF, figures)IComp(TFi, rules_TF, rules_noTF, figures)
TFi |
string or string vector: the transcription factor (or combination of transcription factors) whose importance distribution is evaluated. |
rules_TF |
a set of rules in which |
rules_noTF |
a set of rules obtained from rules_TF removing
the transcription factor (or combination of transcription factors)
in TFi (obtained with the function |
figures |
logical; if |
A list of four elements: the imp element is a vector of
doubles with the importances of TFi in each rule in rulesTF;
the delta element of the list is a list with variations of
distributions of the two measures of support and
confidence. This output is used in the function IPCA
for the Principal Component Analysis of such distributions. The rwi
element is a data.frame with the rules in rulesTF
in which the transcription factor TFi is present
and the rwo element is a data.frame with rules in rwi
obtained removing the transcription factor TFi. Furthermore, if the
input argument figures is set to TRUE, also the plots of the distributions
of support and confidence of the rules before and after removing the
transcription factor TFi are provided.
# Load r_FOSL2 and r_noFOSL2 from the data_man collection of datasets: data('data_man') # The Importance Indexes of FOSL2=1 in the set of rules r_FOSL2 are given by: IComp('FOSL2=1', r_FOSL2, r_noFOSL2, figures=TRUE)# Load r_FOSL2 and r_noFOSL2 from the data_man collection of datasets: data('data_man') # The Importance Indexes of FOSL2=1 in the set of rules r_FOSL2 are given by: IComp('FOSL2=1', r_FOSL2, r_noFOSL2, figures=TRUE)
Within the data_man data collection, the dataset IMP has 4 columns
and 12 rows:
transcription factors present in at least one association rule are listed
in the fist column (TF), the second column (imp) contains the
means of the Importance Index associated with each trascription factor,
the standard deviations for each transcription factor are reported in the
third column (sd) and finally the number of rules found for each
transcription factor are transcribed in the fourth column (nrules).
data("data_man")data("data_man")
An object of class "data.frame"
# IMP is found in the data_man collection of datasets: data("data_man") head(IMP$imp)# IMP is found in the data_man collection of datasets: data("data_man") head(IMP$imp)
Within the data_man data collection, imp_FOSL2 is a list of 4 elements,
containing imp as a numeric vector in which the means of Importance
Indexes are represented, delta as a data.frame in which the
variations of support and confidence measures are reported, rwi as a
data.frame of association rules where FOSL2 is present and rwo as a
data.frame in which the same rules are considered for FOSL2=0.
data("data_man")data("data_man")
An object of class "list"
# imp_FOSL2 is found in the data_man collection of datasets: data("data_man") head(imp_FOSL2$imp)# imp_FOSL2 is found in the data_man collection of datasets: data("data_man") head(imp_FOSL2$imp)
IMP_Z is a list of 12 elements, containig the Importance Indexes for all
transcription factors (p) in the set of relevant association rules
extracted. It is included in the data_man data collection.
data("data_man")data("data_man")
An object of class "list"
# IMP_Z is found in the data_man collection of datasets: data("data_man") head(IMP_Z[[1]])# IMP_Z is found in the data_man collection of datasets: data("data_man") head(IMP_Z[[1]])
The function is used for validation of the defined Importance Index. It is defined as the linear combination of variations of support and confidence measures. The Principal Component Analysis lets the user evaluate if in the 1D reference system defined by such linear combination, it is possible to describe the variability of the data.
IPCA(delta_list, IMP)IPCA(delta_list, IMP)
delta_list |
list of variations of distributions
of support and confidence measures, obtained using the |
IMP |
the importance matrix with the mean Importance Index of every candidate co-regulator transcription factor and the number of rules in which each of them appears. |
Variance explained by every principal component (summary),
scores (i.e., the coordinates) of data in delta_list in the reference system
defined by the principal components (scores) and loadings
(i.e., the coefficinets) of the linear combination that defines
each principal component (loadings).
The plots of the variability, the cumulate percentage of variance explained
by each principal component and loadings of every principal component are
also returned.
# Load IMP, DELTA and TF_Imp from the data_man collection of datasets: data('data_man') colnames(IMP) TF_Imp <- data.frame(IMP$TF, IMP$imp, IMP$nrules) i.pc <- IPCA(DELTA, TF_Imp) names(i.pc)# Load IMP, DELTA and TF_Imp from the data_man collection of datasets: data('data_man') colnames(IMP) TF_Imp <- data.frame(IMP$TF, IMP$imp, IMP$nrules) i.pc <- IPCA(DELTA, TF_Imp) names(i.pc)
The function is used in the elaboration of the left-hand-side of association rules to search for the items.
items(itemset)items(itemset)
itemset |
object of class rules. |
A string vector with the items in the left-hand-side of the rule in itemset.
items('{TAF1=1,EP300=1,MAX=1}') # the output is: 'TAF1=1','EP300=1','MAX=1'items('{TAF1=1,EP300=1,MAX=1}') # the output is: 'TAF1=1','EP300=1','MAX=1'
The function is used in the construction of the left-hand-side of association rules to search for a rule of interest.
itemset(items)itemset(items)
items |
a string vector. |
Itemset of the left-hand-side of a searched rule, with the items
in items.
itemset(c('TAF1=1','EP300=1')) # the output is: 'TAF1=1,EP300=1'itemset(c('TAF1=1','EP300=1')) # the output is: 'TAF1=1,EP300=1'
MCF7_chr1 is a Large GRanges in which metadata columns identify transcription factors and genomic coordinates of regions in the first chromosome of the MCF-7 human breast adenocarcinoma cell line are represented in the left-hand side of the GRanges, therefore each row is a different genomic region.
data("MCF7_chr1")data("MCF7_chr1")
An object of class "GRanges"
data("MCF7_chr1") head(MCF7_chr1)data("MCF7_chr1") head(MCF7_chr1)
p_TFs is a character vector that contains the transcription factors
present in at least one association rule. It is included in the
data_man data collection.
data("data_man")data("data_man")
An object of class "character"
# p_TFs is found in the data_man collection of datasets: data("data_man") p_TFs# p_TFs is found in the data_man collection of datasets: data("data_man") p_TFs
The function is used to find candidate transcription factor co-regulator with a given transcription factor, by finding the transcription factors that contribute to the prediction of the given trascription factor presence in a set of association rules.
presAbs(TFs, rules, type)presAbs(TFs, rules, type)
TFs |
a string vector: set of transcription factors to check their presence (TF=1) or absence (TF=0) in the set of rules. |
rules |
data.frame: rules and their quality measures. |
type |
logical parameter: if |
A list of two string vectors: the list pres contains
all the transcription factors in TFs that are present in rules,
and the list abs contains all the transcription factors in TFs
that are absent in rules.
library(GenomicRanges) # Load r_TEAD4 from the data_man collection of datasets: data('data_man') # Load MCF7_chr1: data('MCF7_chr1') # Transcription factors present in at least one of the regions # in the considered dataset: c <- names(elementMetadata(MCF7_chr1)) names(presAbs(c, r_TEAD4, TRUE)) # Transcription factors present in at least one of the association rules: p_TFs <- presAbs(c, r_TEAD4, TRUE)$pres p_TFslibrary(GenomicRanges) # Load r_TEAD4 from the data_man collection of datasets: data('data_man') # Load MCF7_chr1: data('MCF7_chr1') # Transcription factors present in at least one of the regions # in the considered dataset: c <- names(elementMetadata(MCF7_chr1)) names(presAbs(c, r_TEAD4, TRUE)) # Transcription factors present in at least one of the association rules: p_TFs <- presAbs(c, r_TEAD4, TRUE)$pres p_TFs
Within the data_man data collection, the dataset r_FOSL2 has 5 columns
and 28 rows:
the first column contains the left-hand-side of the rules (lhs),
the second column is the right-hand-side of the rules (rhs),
the third column reports the support measures (support),
the fourth column contains the confidence measures (confidence) and
the lift measures are listed in the fifth column (lift).
data("data_man")data("data_man")
An object of class "data.frame"
# r_FOSL2 is found in the data_man collection of datasets: data("data_man") head(r_FOSL2$lhs)# r_FOSL2 is found in the data_man collection of datasets: data("data_man") head(r_FOSL2$lhs)
Within the data_man data collection, the dataset r_noFOSL2 has 5
columns and 28 rows:
the first column contains the left-hand-side of the rules (lhs),
the second column is the right-hand-side of the rules (rhs),
the third column reports the support measures (support),
the fourth column contains the confidence measures (confidence) and
the lift measures are listed in the fifth column (lift).
data("data_man")data("data_man")
An object of class "data.frame"
# Load r_noFOSL2 from the data_man collection of datasets: data("data_man") head(r_noFOSL2$lhs)# Load r_noFOSL2 from the data_man collection of datasets: data("data_man") head(r_noFOSL2$lhs)
Within the data_man data collection, the dataset r_TEAD4 has 5 columns
and 28 rows:
the first column contains the left-hand-side of the rules (lhs),
the second column is the right-hand-side of the rules (TEAD4=1) (rhs),
the third column reports the support measures (support),
the fourth column contains the confidence measures (confidence) and
the lift measures are listed in the fifth column (lift).
data("data_man")data("data_man")
An object of class "data.frame"
# Load r_TEAD4 from the data_man collection of datasets: data("data_man") head(r_TEAD4$lhs)# Load r_TEAD4 from the data_man collection of datasets: data("data_man") head(r_TEAD4$lhs)
From the dataset in data, the function extracts a set of association
rules, with a certain item in their right-hand-side. Each rule extracted has
support greater than minsupp and confidence greater than
minconf. The extraction is made using the apriori
function implemented in the arules package.
minsupp and minconf thresholds are set by the user in order
to extract a limited number of most relevant association rules.
rulesGen(data, TF, minsupp, minconf, type)rulesGen(data, TF, minsupp, minconf, type)
data |
a GRanges object in which the metadata columns contain the Indicator of presence matrix i.e., a matrix with 1 and 0 values representing presence or absence, respectively (in case other values different from 0 are present, all of them are considered as representing presence). |
TF |
a string with the name of the trancription factor wanted in the right-hand-side of the extracted rules. |
minsupp |
an integer, the minimal support of the extracted rules. |
minconf |
an integer, the minimal confidence of the extracted rules. |
type |
a logical parameter; if |
A data frame with the association rules extracted and their quality measures of support, confidence and lift.
# Load the dataset: data('MCF7_chr1') # To extract association rules from data, with TEAD4=1 in the right-hand-side # and support greater than 0.005 and confidence greater than 0.62: # r_TEAD4 <- rulesGen(data, 'TEAD4=1', minsupp=0.005, minconf=0.62, # type=TRUE) r_TEAD4 <- rulesGen(MCF7_chr1, 'TEAD4=1', 0.005, 0.62, TRUE)# Load the dataset: data('MCF7_chr1') # To extract association rules from data, with TEAD4=1 in the right-hand-side # and support greater than 0.005 and confidence greater than 0.62: # r_TEAD4 <- rulesGen(data, 'TEAD4=1', minsupp=0.005, minconf=0.62, # type=TRUE) r_TEAD4 <- rulesGen(MCF7_chr1, 'TEAD4=1', 0.005, 0.62, TRUE)
From a set of relevant association rules, only the ones containing TFi
in their left-hand-side are subsetted, together with their quality measures
of support, confidence and lift. The function is then used for the evaluation
of the importance distribution of TFi.
rulesTF(TFi, rules, verbose)rulesTF(TFi, rules, verbose)
TFi |
a string with the name of the transcription factor (or combination of transcription factors) wanted in the left-hand-side of the rules to find. |
rules |
a data.frame with association rules and their quality measures of support, confidence and lift. |
verbose |
a logical parameter. If |
A data.frame with association rules containing TFi in their
left-hand-side, with their quality measures of support, confidence and lift.
# Load r_TEAD4 from the data_man collection of datasets: data('data_man') r_FOSL2 <- rulesTF('FOSL2=1', r_TEAD4, verbose=FALSE)# Load r_TEAD4 from the data_man collection of datasets: data('data_man') r_FOSL2 <- rulesTF('FOSL2=1', r_TEAD4, verbose=FALSE)
The function substitutes the presence of a given transcription factor
TFi (or a combination of transcription factors) chosen by the
user with its absence, in the subset of relevant association rules
extracted with the function rulesTF. Then it searches
for the obtained rules and their quality measures of
support, confidence and lift in the set of most relevant associations
extracted with the function rulesGen.
A rule is searched in all the association
rules generable from the considered dataset using the function
search_rule.
rulesTF0(TFi, sub_rules, all_rules, data, RHS)rulesTF0(TFi, sub_rules, all_rules, data, RHS)
TFi |
a string, or a string vector: transcription factor (or combination of transcription factors) to remove from the set of rules. |
sub_rules |
a data.frame with a subset of rules containing |
all_rules |
a data.frame with a set of all the rules and their quality measures of support, confidence and lift, to be considered for the search of the obtained rules and their quality measures. |
data |
a GRanges object which contains the Indicator of presence matrix i.e., a matrix with 1 and 0 values representing presence or absence, respectively (in case other values different from 0 are present, all of them are considered as representing presence). |
RHS |
the right-hand-side of the considered association rules. |
A data.frame with all the rules in the set sub_rules
in which the transcription factor (or combination of transcription factors)
TFi is absent, and their quality measures of support, confidence
and lift.
# Load r_TEAD4 and r_FOSL2 from the data_man collection of datasets: data("data_man") # Load MCF7_chr1: data("MCF7_chr1") r_noFOSL2 <- rulesTF0("FOSL2=1", r_FOSL2, r_TEAD4, MCF7_chr1, "TEAD4=1")# Load r_TEAD4 and r_FOSL2 from the data_man collection of datasets: data("data_man") # Load MCF7_chr1: data("MCF7_chr1") r_noFOSL2 <- rulesTF0("FOSL2=1", r_FOSL2, r_TEAD4, MCF7_chr1, "TEAD4=1")
The function looks for an association rule in a dataset data
with a fixed left-hand-side (LHS) and a fixed right-hand-side
(RHS). The function is used in the rulesTF0.
search_rule(data, LHS, RHS)search_rule(data, LHS, RHS)
data |
a binary matrix or data.frame in which the rule is searched |
LHS |
a string with the left-hand-side of the searched rule |
RHS |
a string with the right-hand-side of the searched rule |
A vector with five elements: the left-hand-side of the searched rule, the right-hand-side of the searched rule, the support, confidence and lift of the found rule.
Within the data_man data collection, the dataset TF_Imp has 3 columns
and 12 rows:
the fist column contains the transcription factors (IMP.TF),
the Importance Indexes associated with each trascription factor
are listed in the second column (IMP.imp) and the third column
contains the number of rules found for each transcription factor
(IMP.nrules).
data("data_man")data("data_man")
An object of class "data.frame"
# TF_Imp is found in the data_man collection of datasets: data("data_man") head(TF_Imp$IMP.TF)# TF_Imp is found in the data_man collection of datasets: data("data_man") head(TF_Imp$IMP.TF)