tests/testthat/test-RVConsensusOPLS.R

In ConsensusOPLS: Consensus OPLS for Multi-Block Data Fusion

test_that("RVConsensusOPLS", {
    ## rvcopls output is compared to Matlab output
    rvcopls <- RVConsensusOPLS(
        data=demo_3_Omics[c("MetaboData", "MicroData", "ProteoData")],
        Y=demo_3_Omics$Y,
        maxOcomp=10,
        nfold=14)
    
    ## test functionality
    rvcopls.fac <- RVConsensusOPLS(
        data=demo_3_Omics[c("MetaboData", "MicroData", "ProteoData")],
        Y=factor(c(rep('M',7), rep('F',7))),
        nfold=3)
    rvcopls.reg <- RVConsensusOPLS(
        data=demo_3_Omics[c("MetaboData", "MicroData", "ProteoData")],
        Y=c(rnorm(7, mean=1, sd=0.01), rnorm(7, mean=0, sd=0.01)),
        modelType="reg",
        nfold=3)
    
    ## RV
    expect_equal(rvcopls$RV,
                 c(MetaboData=0.770310488895171,
                   MicroData=0.835375809822633,
                   ProteoData=0.74971449858494),
                 tolerance=1e-6)
    
    ## lambda
    expect_equal(rvcopls$koplsModel$lambda[,1], 
                 c(MetaboData=0.0455910270422634,
                   MicroData=0.00474677906005336,
                   ProteoData=0.00357785429350788),
                 tolerance=1e-6)
    expect_equal(rvcopls$koplsModel$lambda[,2], 
                 c(MetaboData=0.236514988519443,
                   MicroData=0.0231772225562188,
                   ProteoData=0.153126721989076),
                 tolerance=1e-6)
    
    ## blockContribution
    expect_equal(rvcopls$koplsModel$blockContribution[,1], 
                 c(MetaboData=0.84559897268353,
                   MicroData=0.0880408220024505,
                   ProteoData=0.0663602053140197),
                 tolerance=1e-6)
    expect_equal(rvcopls$koplsModel$blockContribution[,2], 
                 c(MetaboData=0.572926698791582,
                   MicroData=0.0561437974371785,
                   ProteoData=0.37092950377124),
                 tolerance=1e-6)
    
    ## normKernels
    expect_equal(rvcopls$normKernels[[1]][1,2],
                 c(0.0328819715926973),
                 tolerance=1e-6)
    expect_equal(rvcopls$normKernels[[2]][7,7],
                 c(0.104945253831102),
                 tolerance=1e-6)
    expect_equal(rvcopls$normKernels[[3]][9,11],
                 c(0.121529241109287),
                 tolerance=1e-6)
    
    #### Optimal model
    ## Kpreproc
    expect_equal(rvcopls$koplsModel$K[1,1][[1]][9,11],
                 c(0.006208603),
                 tolerance=1e-6)
    expect_equal(rvcopls$koplsModel$K[1,1][[1]][3,13],
                 c(-0.01054343),
                 tolerance=1e-6)
    
    ## Cp
    expect_equal(unname(rvcopls$koplsModel$Cp[,1]),
                 c(-0.707106781186547,
                    0.707106781186548), tolerance = 1e-6)
    
    ## Sp
    expect_equal(1/rvcopls$koplsModel$Sp[1,1]^2,
                 c(0.9574551557441),
                 tolerance=1e-6)
    
    ## Up
    expect_equal(unique(rvcopls$koplsModel$Up[,1]),
                c(-0.707106781186547,
                  0.707106781186547),
                tolerance=1e-6)
    
    ## params
    expect_equal(rvcopls$koplsModel$params$nOcomp,
                 c(1))
    
    ## Tp
    expect_equal(rvcopls$koplsModel$Tp[[1]][1:3],
                 c(0.0942610932887749,
                   -0.154506780676698,
                   -0.0608792282254192),
                 tolerance=1e-6)
    expect_equal(rvcopls$koplsModel$Tp[[2]][4:6],
                 c(-0.00916472674616893,
                   -0.208638510110777,
                   -0.115250463686483),
                 tolerance=1e-6)
    
    ## scores
    expect_equal(unname(rvcopls$scores[6:8,'p_1']),
                 c(-0.115250463686483,
                   -0.134547425243546,
                   0.169584736593916),
                 tolerance=1e-6)
    expect_equal(unname(rvcopls$scores[6:8,'o_1']),
                 c(0.225639604453228,
                   -0.750090705302492,
                   -0.246527124390688),
                 tolerance=1e-6)
                 
    ## loadings
    expect_equal(rvcopls$loadings$MetaboData[1:3,'p_1'],
                 c(MT15708=-3.7904893970217,
                   MT15709=0.477951498442441,
                   MT15710=0.597346765073993),
                 tolerance=1e-6)
    expect_equal(rvcopls$loadings$MetaboData[1:3,'o_1'],
                 c(MT15708=-0.0906993946312754,
                   MT15709=0.391637521782826,
                   MT15710=-0.102030862736768),
                 tolerance=1e-6)
    
    expect_equal(rvcopls$loadings$MicroData[1:3,'p_1'],
                 c(GC26855=-121.851575745773,
                   GC26856=132.585031777677,
                   GC26857=75.1435270079714),
                 tolerance=1e-6)
    expect_equal(rvcopls$loadings$MicroData[1:3,'o_1'],
                 c(GC26855=-28.2239137515635,
                   GC26856=-31.9262881875121,
                   GC26857=1.89187255025149),
                 tolerance=1e-6)
    
    expect_equal(rvcopls$loadings$ProteoData[1:3,'p_1'],
                 c(MT1126=-8838.55937473373,
                   MT1127=1072.45583338495,
                   MT1128=-232.657268097895),
                 tolerance=1e-6)
    expect_equal(rvcopls$loadings$ProteoData[1:3,'o_1'],
                 c(MT1126=162.821016719371,
                   MT1127=-41.4788108974671,
                   MT1128=-18.1358103752954),
                 tolerance=1e-6)
    
    #### CV
    ## AllYhat
    expect_equal(unname(rvcopls$cv$AllYhat[1:3,1]),
                 c(0.1381898,
                   0.6109852,
                   0.5683056),
                 tolerance=1e-6)
    expect_equal(unname(rvcopls$cv$AllYhat[4:6,2]),
                 c(0.60339737,
                   0.23570761,
                   0.48595049),
                 tolerance=1e-6)
    expect_equal(unname(rvcopls$cv$AllYhat[1:3,3]),
                 c(0.136238770,
                   0.524588637,
                   0.740291299),
                 tolerance=1e-6)
    expect_equal(unname(rvcopls$cv$AllYhat[7:9,4]),
                 c(-0.2020121,
                   0.9947210,
                   0.4679297),
                 tolerance=1e-6)
    
    ## Q2Yhat
    expect_equal(unname(rvcopls$cv$Q2Yhat[1:5]),
                 c(0.24144459290868,
                   0.254116314089248,
                   0.0688378757417817,
                   -0.058455510451356,
                   0.159444129092519),
                 tolerance=1e-6)
    
    ## DQ2Yhat
    expect_equal(unname(rvcopls$cv$DQ2Yhat[1:5]),
                 c(0.120292310258979,
                   0.151671627027632,
                   -0.0418685758525927,
                   -0.171662986447783,
                   0.0788120118584003),
                 tolerance=1e-6)
    
    # ## Tcv
    # expect_equal(rvcopls$cv$Tcv[6:8,1],
    #              c(-3.05298886630587,
    #                -10.7977788716645,
    #                2.7915511228586),
    #              tolerance=1e-6)
    # 
    # ## Yhat (at maxOcomp)
    # expect_equal(rvcopls$cv$Yhat[6:8,1],
    #              c(0.80176823634328,
    #                1.56729072041118,
    #                0.224073196497945),
    #              tolerance=1e-6)
    # expect_equal(rvcopls$cv$Yhat[6:8,2],
    #              c(0.19823176365672,
    #                -0.567290720411185,
    #                0.775926803502055),
    #              tolerance=1e-6)
    # 
    # expect_equal(rvcopls$koplsModel$R2X,
    #              c(0.18253923588794,
    #                0.257726827483885),
    #              tolerance=1e-6)
    # expect_equal(rvcopls$koplsModel$R2XO,
    #              c(0,0.0811103596697501),
    #              tolerance=1e-6)
    # expect_equal(rvcopls$koplsModel$R2XC,
    #              c(0.18253923588794,
    #                0.176616467814135),
    #              tolerance=1e-6)
})