Exercice 2

Load data

• Load the 2 .Rdata files that you can find in the data/ directory. It contains gene expression data and metadata about experimental design and 3 mouse strains used (B6, D2, AK)
• Using ls() you can find the variables saved in these files
• Which type of variables are they ?

Show answer !

First we load the data:

setwd("data")
load("Ex2.expression.Rdata")
load("Ex2.metadata.Rdata")

We then list all variables present in our environment:

ls()

## [1] "expression" "metadata"

We see which type of variable are expression and metadata

class(expression)

## [1] "data.frame"

class(metadata)

## [1] "data.frame"

Both of them are data.frame

Inspect data

• Have a look at the data
• Describe them
• Can you see some problems using the summary() function ?

Show answer !

We can easly see that the order of sample ID in the metadata (rows) and in the expression data (columns) are not the same order. It can be problematic if we write some statistical test in R.

metadata

expression

Using summary, we don’t see obvious problems. Strain, Condition and Tissue are in charactet, we should transform them into factors to maybe detect some typos

summary(metadata)

##       ID             SampleName             Time           ZTime       
##  Length:71          Length:71          Min.   : 0.00   Min.   : 0.000  
##  Class :character   Class :character   1st Qu.: 9.00   1st Qu.: 3.000  
##  Mode  :character   Mode  :character   Median :18.00   Median : 6.000  
##                                        Mean   :23.75   Mean   : 8.873  
##                                        3rd Qu.:36.00   3rd Qu.:12.000  
##                                        Max.   :48.00   Max.   :18.000  
##   Condition            Tissu              Strain         
##  Length:71          Length:71          Length:71         
##  Class :character   Class :character   Class :character  
##  Mode  :character   Mode  :character   Mode  :character  
##                                                          
##                                                          
## 

Having a quick look at expression data, we see that GSM239937 and GSM239879 are in character type while they should be numeric like the other

summary(expression,digits = 2,quantile.type=2)

##    GSM239929      GSM239904      GSM239898      GSM239901      GSM239873   
##  Min.   : 3.5   Min.   : 3.5   Min.   : 3.4   Min.   : 3.4   Min.   : 3.4  
##  1st Qu.: 5.2   1st Qu.: 5.3   1st Qu.: 5.2   1st Qu.: 5.3   1st Qu.: 5.3  
##  Median : 6.7   Median : 7.0   Median : 7.0   Median : 7.0   Median : 7.1  
##  Mean   : 6.8   Mean   : 6.9   Mean   : 7.0   Mean   : 6.9   Mean   : 7.0  
##  3rd Qu.: 8.1   3rd Qu.: 8.6   3rd Qu.: 8.6   3rd Qu.: 8.6   3rd Qu.: 8.3  
##  Max.   :11.2   Max.   :11.4   Max.   :11.4   Max.   :11.3   Max.   :11.2  
##    GSM239934      GSM239887      GSM239927      GSM239882     GSM239935        
##  Min.   : 3.4   Min.   : 3.4   Min.   : 3.3   Min.   : 3.4   Length:100        
##  1st Qu.: 5.1   1st Qu.: 5.4   1st Qu.: 5.1   1st Qu.: 5.4   Class :character  
##  Median : 6.9   Median : 7.0   Median : 6.9   Median : 6.9   Mode  :character  
##  Mean   : 6.8   Mean   : 7.0   Mean   : 6.8   Mean   : 7.0                     
##  3rd Qu.: 8.1   3rd Qu.: 8.4   3rd Qu.: 8.1   3rd Qu.: 8.3                     
##  Max.   :11.2   Max.   :11.3   Max.   :11.2   Max.   :11.2                     
##    GSM239909      GSM239890      GSM239879      GSM239893      GSM239908   
##  Min.   : 3.4   Min.   : 3.2   Min.   : 3.4   Min.   : 3.5   Min.   : 3.4  
##  1st Qu.: 5.3   1st Qu.: 5.5   1st Qu.: 5.3   1st Qu.: 5.3   1st Qu.: 5.3  
##  Median : 7.0   Median : 7.0   Median : 7.1   Median : 7.0   Median : 7.0  
##  Mean   : 6.9   Mean   : 7.0   Mean   : 7.0   Mean   : 6.9   Mean   : 7.0  
##  3rd Qu.: 8.6   3rd Qu.: 8.6   3rd Qu.: 8.4   3rd Qu.: 8.6   3rd Qu.: 8.6  
##  Max.   :11.3   Max.   :11.3   Max.   :11.2   Max.   :11.2   Max.   :11.1  
##    GSM239868      GSM239869      GSM239891      GSM239928     GSM239892        
##  Min.   : 3.2   Min.   : 3.5   Min.   : 3.5   Min.   : 3.5   Length:100        
##  1st Qu.: 5.4   1st Qu.: 5.4   1st Qu.: 5.3   1st Qu.: 5.1   Class :character  
##  Median : 6.9   Median : 7.0   Median : 7.0   Median : 6.8   Mode  :character  
##  Mean   : 7.0   Mean   : 7.0   Mean   : 6.9   Mean   : 6.8                     
##  3rd Qu.: 8.5   3rd Qu.: 8.5   3rd Qu.: 8.6   3rd Qu.: 8.1                     
##  Max.   :11.3   Max.   :11.3   Max.   :11.3   Max.   :11.3                     
##    GSM239895      GSM239910      GSM239881      GSM239919      GSM239921   
##  Min.   : 3.5   Min.   : 3.5   Min.   : 3.4   Min.   : 3.3   Min.   : 3.4  
##  1st Qu.: 5.3   1st Qu.: 5.3   1st Qu.: 5.3   1st Qu.: 5.1   1st Qu.: 5.0  
##  Median : 7.0   Median : 7.0   Median : 7.0   Median : 6.8   Median : 6.6  
##  Mean   : 7.0   Mean   : 6.9   Mean   : 7.0   Mean   : 6.8   Mean   : 6.8  
##  3rd Qu.: 8.6   3rd Qu.: 8.6   3rd Qu.: 8.6   3rd Qu.: 8.0   3rd Qu.: 8.0  
##  Max.   :11.3   Max.   :11.2   Max.   :11.3   Max.   :11.3   Max.   :11.2  
##    GSM239936      GSM239899      GSM239894      GSM239877      GSM239917   
##  Min.   : 3.2   Min.   : 3.2   Min.   : 3.3   Min.   : 3.4   Min.   : 3.5  
##  1st Qu.: 5.2   1st Qu.: 5.3   1st Qu.: 5.3   1st Qu.: 5.4   1st Qu.: 5.2  
##  Median : 6.9   Median : 7.0   Median : 7.0   Median : 7.0   Median : 6.6  
##  Mean   : 6.8   Mean   : 7.0   Mean   : 7.0   Mean   : 7.0   Mean   : 6.8  
##  3rd Qu.: 8.2   3rd Qu.: 8.6   3rd Qu.: 8.6   3rd Qu.: 8.5   3rd Qu.: 8.0  
##  Max.   :11.3   Max.   :11.3   Max.   :11.3   Max.   :11.2   Max.   :11.2  
##    GSM239937      GSM239930      GSM239872      GSM239902      GSM239915   
##  Min.   : 3.4   Min.   : 3.4   Min.   : 3.3   Min.   : 3.5   Min.   : 3.3  
##  1st Qu.: 5.2   1st Qu.: 4.8   1st Qu.: 5.5   1st Qu.: 5.4   1st Qu.: 5.1  
##  Median : 6.8   Median : 6.8   Median : 7.0   Median : 7.0   Median : 6.9  
##  Mean   : 6.8   Mean   : 6.8   Mean   : 7.0   Mean   : 7.0   Mean   : 6.8  
##  3rd Qu.: 8.1   3rd Qu.: 8.1   3rd Qu.: 8.4   3rd Qu.: 8.5   3rd Qu.: 8.1  
##  Max.   :11.2   Max.   :11.2   Max.   :11.3   Max.   :11.3   Max.   :11.3  
##    GSM239903      GSM239923      GSM239900      GSM239933      GSM239870   
##  Min.   : 3.4   Min.   : 3.5   Min.   : 3.6   Min.   : 3.5   Min.   : 3.3  
##  1st Qu.: 5.2   1st Qu.: 5.1   1st Qu.: 5.3   1st Qu.: 5.1   1st Qu.: 5.3  
##  Median : 7.0   Median : 6.8   Median : 7.0   Median : 6.8   Median : 7.0  
##  Mean   : 7.0   Mean   : 6.8   Mean   : 6.9   Mean   : 6.8   Mean   : 7.0  
##  3rd Qu.: 8.6   3rd Qu.: 8.0   3rd Qu.: 8.6   3rd Qu.: 8.1   3rd Qu.: 8.4  
##  Max.   :11.3   Max.   :11.3   Max.   :11.3   Max.   :11.2   Max.   :11.0  
##    GSM239889      GSM239924      GSM239880      GSM239878      GSM239931   
##  Min.   : 3.5   Min.   : 3.4   Min.   : 3.2   Min.   : 3.6   Min.   : 3.3  
##  1st Qu.: 5.4   1st Qu.: 5.1   1st Qu.: 5.4   1st Qu.: 5.3   1st Qu.: 5.1  
##  Median : 7.0   Median : 6.8   Median : 7.0   Median : 7.0   Median : 6.9  
##  Mean   : 7.0   Mean   : 6.8   Mean   : 7.0   Mean   : 7.0   Mean   : 6.8  
##  3rd Qu.: 8.5   3rd Qu.: 8.1   3rd Qu.: 8.4   3rd Qu.: 8.4   3rd Qu.: 8.1  
##  Max.   :11.2   Max.   :11.2   Max.   :11.3   Max.   :11.3   Max.   :11.1  
##    GSM239884      GSM239896      GSM239871      GSM239920      GSM239916   
##  Min.   : 3.4   Min.   : 3.5   Min.   : 3.5   Min.   : 3.4   Min.   : 3.3  
##  1st Qu.: 5.4   1st Qu.: 5.3   1st Qu.: 5.3   1st Qu.: 5.0   1st Qu.: 5.2  
##  Median : 6.9   Median : 7.0   Median : 7.0   Median : 6.7   Median : 6.8  
##  Mean   : 7.0   Mean   : 7.0   Mean   : 7.0   Mean   : 6.8   Mean   : 6.9  
##  3rd Qu.: 8.5   3rd Qu.: 8.6   3rd Qu.: 8.3   3rd Qu.: 7.9   3rd Qu.: 8.1  
##  Max.   :11.2   Max.   :11.4   Max.   :11.3   Max.   :11.3   Max.   :11.3  
##    GSM239918      GSM239876      GSM239897      GSM239914      GSM239938   
##  Min.   : 3.4   Min.   : 3.6   Min.   : 3.5   Min.   : 3.5   Min.   : 3.3  
##  1st Qu.: 5.0   1st Qu.: 5.3   1st Qu.: 5.4   1st Qu.: 5.3   1st Qu.: 5.1  
##  Median : 6.6   Median : 7.0   Median : 7.0   Median : 7.0   Median : 6.8  
##  Mean   : 6.8   Mean   : 7.0   Mean   : 7.0   Mean   : 7.0   Mean   : 6.8  
##  3rd Qu.: 8.1   3rd Qu.: 8.4   3rd Qu.: 8.5   3rd Qu.: 8.7   3rd Qu.: 8.1  
##  Max.   :11.3   Max.   :11.4   Max.   :11.4   Max.   :11.3   Max.   :11.2  
##    GSM239907      GSM239925      GSM239922      GSM239926      GSM239885   
##  Min.   : 3.3   Min.   : 3.5   Min.   : 3.3   Min.   : 3.2   Min.   : 3.5  
##  1st Qu.: 5.2   1st Qu.: 5.1   1st Qu.: 5.1   1st Qu.: 5.0   1st Qu.: 5.3  
##  Median : 7.0   Median : 6.9   Median : 6.8   Median : 7.0   Median : 7.0  
##  Mean   : 6.9   Mean   : 6.8   Mean   : 6.8   Mean   : 6.8   Mean   : 7.0  
##  3rd Qu.: 8.5   3rd Qu.: 8.1   3rd Qu.: 8.0   3rd Qu.: 8.1   3rd Qu.: 8.4  
##  Max.   :11.3   Max.   :11.2   Max.   :11.3   Max.   :11.2   Max.   :11.2  
##    GSM239888      GSM239906      GSM239883      GSM239913      GSM239912   
##  Min.   : 3.2   Min.   : 3.1   Min.   : 3.4   Min.   : 3.4   Min.   : 3.6  
##  1st Qu.: 5.4   1st Qu.: 5.2   1st Qu.: 5.3   1st Qu.: 5.3   1st Qu.: 5.4  
##  Median : 7.0   Median : 7.1   Median : 7.0   Median : 7.0   Median : 7.0  
##  Mean   : 7.0   Mean   : 6.9   Mean   : 7.0   Mean   : 6.9   Mean   : 7.0  
##  3rd Qu.: 8.4   3rd Qu.: 8.6   3rd Qu.: 8.5   3rd Qu.: 8.7   3rd Qu.: 8.6  
##  Max.   :11.2   Max.   :11.2   Max.   :11.2   Max.   :11.3   Max.   :11.2  
##    GSM239911      GSM239886      GSM239875      GSM239874      GSM239932   
##  Min.   : 3.4   Min.   : 3.3   Min.   : 3.2   Min.   : 3.6   Min.   : 3.5  
##  1st Qu.: 5.2   1st Qu.: 5.4   1st Qu.: 5.4   1st Qu.: 5.4   1st Qu.: 5.0  
##  Median : 7.2   Median : 7.0   Median : 7.0   Median : 7.0   Median : 6.9  
##  Mean   : 7.0   Mean   : 7.0   Mean   : 7.0   Mean   : 7.0   Mean   : 6.8  
##  3rd Qu.: 8.7   3rd Qu.: 8.4   3rd Qu.: 8.3   3rd Qu.: 8.2   3rd Qu.: 8.0  
##  Max.   :11.2   Max.   :11.2   Max.   :11.4   Max.   :11.3   Max.   :11.2

Reformat

• Reformat the metadata and expression data that are needed
• Do you see new problems with the summary() function ?

Show answer !

We can transform some information in the metadata into factors:

metadata$Strain<-as.factor(metadata$Strain)
metadata$Condition<-as.factor(metadata$Condition)
metadata$Tissu<-as.factor(metadata$Tissu)

Using a summary, we see better potential problems:

summary(metadata)

##       ID             SampleName             Time           ZTime       
##  Length:71          Length:71          Min.   : 0.00   Min.   : 0.000  
##  Class :character   Class :character   1st Qu.: 9.00   1st Qu.: 3.000  
##  Mode  :character   Mode  :character   Median :18.00   Median : 6.000  
##                                        Mean   :23.75   Mean   : 8.873  
##                                        3rd Qu.:36.00   3rd Qu.:12.000  
##                                        Max.   :48.00   Max.   :18.000  
##  Condition   Tissu     Strain  
##  Ctr:36    Brain:71   AK  :23  
##  SD :35               B6  :22  
##                       d2  : 1  
##                       D2  :23  
##                       NaN : 1  
##                       NA's: 1

Some samples are unknown mouse Strain, they will be removed later and we see a typo. One mouse was annotated “d2” instead of “D2”.

We have to rename it:

metadata$Strain[which(metadata$Strain ==  "d2")]<-"D2"

For the experession data, we can see the 2 problematic recordings:

expression$GSM239892

##   [1] "7.1516016314504"  "9.44083982847215" "5.58214299030138"
##   [4] "8.77636374861736" "NA"               "7.01073848225408"
##   [7] "7.51168547926991" "10.9942396307951" "3.5426477506181" 
##  [10] "11.2689940905563" "4.53300849132838" "3.93627124437631"
##  [13] "4.22529429703503" "9.64249266538495" "4.75981844248776"
##  [16] "7.90611853773516" "8.43996954444962" "9.31335735689715"
##  [19] "7.21183422219714" "9.43023479379811" "7.56275579853497"
##  [22] "6.2637628719418"  "4.55087222927257" "6.36221424813594"
##  [25] "5.30887673968548" "6.65887027754669" "3.98358500197667"
##  [28] "5.63816902313448" "6.54750271986372" "3.76705969902158"
##  [31] "7.13752470247227" "6.60720091584353" "6.86501210504626"
##  [34] "6.09345185819473" "3.78261561756992" "10.3716916857919"
##  [37] "9.8656180529183"  "7.88514755796799" "4.68059094256405"
##  [40] "7.20394953277159" "9.48601693614409" "3.58116580206789"
##  [43] "8.40294133877664" "3.91343915415062" "6.17218208925419"
##  [46] "4.71936252106112" "7.84634257411119" "7.07133973499474"
##  [49] "7.48043881913625" "5.38381073453594" "6.4117148219353" 
##  [52] "10.8638181809604" "8.53203369897089" "9.71412964400098"
##  [55] "4.41553452782242" "7.52919043197241" "9.24428105566894"
##  [58] "7.76586241594676" "3.48750361484482" "5.42885051906792"
##  [61] "5.98329549404109" "3.57968102404718" "6.94199154485471"
##  [64] "7.76108577095632" "4.41344177163489" "6.44354872772903"
##  [67] "5.41883662252547" "7.59619062955516" "4.81955276503251"
##  [70] "7.59033448049269" "7.35294344938346" "5.02283676418099"
##  [73] "8.03051275657304" "3.99357055438367" "4.39597142665914"
##  [76] "5.53312866101338" "9.04157200655132" "6.33434060568062"
##  [79] "7.60263538244219" "4.74136709397252" "6.20835220106041"
##  [82] "10.2598337137952" "8.61491530308567" "5.24883741386779"
##  [85] "7.64891616685063" "7.17972217549752" "7.43210738888375"
##  [88] "10.2739252040157" "4.96486634234627" "8.46040562013735"
##  [91] "8.85002885322223" "5.33339849536578" "10.1298644723963"
##  [94] "10.7028549569907" "9.28967262285694" "5.43844387208594"
##  [97] "9.93303086975839" "10.2353270381247" "7.43120585339832"
## [100] "8.13002510919751"

expression$GSM239935

##   [1] "7.05629185690423" "NA"               "3.62109088251407"
##   [4] "8.707745577559"   "5.39374970472041" "7.08153511654968"
##   [7] "7.74787747159035" "10.9680636213069" "3.80060486399479"
##  [10] "11.1107013306742" "4.71197277197259" "3.53585641389607"
##  [13] "4.01707996453672" "9.71463446247705" "4.6415635029193" 
##  [16] "7.78199353116428" "8.46257287969743" "7.73506505192023"
##  [19] "4.09695801725552" "9.15576828743918" "7.35418669669635"
##  [22] "6.22118577527544" "4.44705855050427" "10.4432610264594"
##  [25] "5.35815577184569" "6.52081278399767" "3.92886780339795"
##  [28] "5.73817284791818" "4.60829120282569" "3.63324879156339"
##  [31] "8.47825705219818" "5.91755868384873" "7.102553765068"  
##  [34] "5.62531797842723" "3.90491726990032" "8.96141708728896"
##  [37] "10.0578429941317" "5.05643766900686" "4.72305652738904"
##  [40] "5.27180439729869" "9.85817578542504" "3.67150002114846"
##  [43] "8.56246753514483" "3.77349036449541" "6.23681159172155"
##  [46] "4.71184400723521" "8.06911097400442" "7.32468290670881"
##  [49] "7.47600606461747" "5.38473025175546" "6.14774027107141"
##  [52] "10.8409328004626" "8.92639609384679" "9.77784799335737"
##  [55] "4.57294092686079" "7.73700514863385" "9.18907529161975"
##  [58] "7.49225626759818" "3.62792684046304" "7.81764997988425"
##  [61] "5.83829787733053" "6.40617750266875" "6.82146934203883"
##  [64] "7.68448055830424" "4.37590105568283" "6.3576105661357" 
##  [67] "6.08551924346881" "7.74021184367948" "4.69291810629799"
##  [70] "7.54517212525693" "7.28068017269519" "5.00653506646486"
##  [73] "7.9484165215413"  "3.95159948552454" "8.81059463856966"
##  [76] "5.57324942755916" "6.18176633687667" "6.34246068904597"
##  [79] "7.49337681197434" "5.00314068374953" "5.99878340297803"
##  [82] "10.0377220337655" "10.4104735822674" "5.13086146584666"
##  [85] "7.82249697648517" "6.74553140259875" "7.06255159524849"
##  [88] "10.1775570310208" "5.11567778914582" "6.85919004528415"
##  [91] "10.4412623020071" "5.43434853189694" "7.49498210719405"
##  [94] "10.4708706063564" "4.70574696727817" "5.76291643609249"
##  [97] "8.06910207153329" "9.87247755896656" "7.37119130778341"
## [100] "6.32179850070519"

We see that the “NA” caused data to be saved as character, we can simply change them into numerical value:

expression$GSM239892<-as.numeric(expression$GSM239892)

## Warning: NAs introduits lors de la conversion automatique

expression$GSM239935<-as.numeric(expression$GSM239935)

## Warning: NAs introduits lors de la conversion automatique

We can save our data into a matrix, if all data are numerical it should be transformed into a numerical matrix.

expression<-as.matrix(expression)

All values are numerical now

typeof(expression)

## [1] "double"

Filter metadata

• Only keep the following strains: B6, D2 in your metadata

Show answer !

With the %in% operator, we ask which strain are B6 and D2 and only keep row which are TRUE

idx<-metadata$Strain %in% c("B6","D2")
idx

##  [1] FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE
## [13] FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE  TRUE
## [25]  TRUE  TRUE  TRUE  TRUE  TRUE FALSE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE
## [37]  TRUE  TRUE  TRUE FALSE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE
## [49]  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE
## [61]  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE  TRUE

metadata<-metadata[idx,]

Reorder

• Filter accordingly the expression data.
• Metadata Samples ID and expression data samples should be in the same order

Show answer !

Directly reordering expression does not work, because 1 sample is missing

expression<-expression[,metadata$ID]

Keep only shared samples

metadata<-metadata[metadata$ID %in% colnames(expression),]
expression<-expression[,metadata$ID]

We can do a control that they are in the same order:

table(colnames(expression) == metadata$ID)

## 
## TRUE 
##   45

Filter

• Filter low expressed genes with a mean less than 5

Show answer !

Some row contain NA value, we have to use the option na.rm=T to avoid them

expression<-expression[rowMeans(expression,na.rm=T)>5,]
dim(expression)

## [1] 79 45

NA remove

• Remove all genes with empty gene expression value

Show answer !

With the function is.na() we can detect which value are NA. the option arr.ind =T will return the row and column with na.

which(is.na(expression),arr.ind=T)

##         row col
## Ddx31     4   2
## Zc2hc1a   2  42

The gene on row 1 contains an empty value, we remove it.

expression<-expression[-1,]

Plot

• Plot the mean expression overall gene using plot()

Show answer !

plot(rowMeans(expression))

Save data

• Save metadata and expression data using the following command:

save(metadata,expression,file="data/Ex2.Results.Rdata")