Skip to main content

Table 1 Performance comparison of different integrative analyses on simulated datasets. The average of each metric was presented and the standard deviation was not shown because the metric values are very stable between different numerical repeats

From: Integrating heterogeneous genomic data to accurately identify disease subtypes

Scale issue Data1 Data2 Concatenation SNF iBFE1 iBFE2 iBFE2
PCCintraclass 0.023 ± 0.0011 0.025 ± 0.0019 0.031 ± 0.0034 0.17 ± 0.0067 0.30 ± 0.012 0.23 ± 0.014 0.28 ± 0.015
PCCinterclass −0.0018 ± 0.00021 −0.0067 ± 0.00033 0.0068 ± 0.00035 0.15 ± 0.0054 −0.30 ± 0.013 −0.21 ± 0.015 −0.26 ± 0.015
PCCintraclass-PCCinterclass 0.024 ± 0.0014 0.031 ± 0.0020 0.024 ± 0.0035 0.021 ± 0.012 0.60 ± 0.022 0.44 ± 0.024 0.54 ± 0.025
Simintraclass 0.02 ± 0.0023 0.02 ± 0.0022 0.02 ± 0.0023 0.97 ± 0.004 0.32 ± 0.011 0.23 ± 0.012 0.29 ± 0.010
Siminterclass 0 ± 0.0 0 ± 0.0 0 ± 0.0 0.97 ± 0.005 0.22 ± 0.012 0.17 ± 0.013 0.22 ± 0.013
Simintraclass-Siminterclass 0.02 ± 0.0023 0.02 ± 0.0022 0.02 ± 0.0023 0.00057 ± 0.0011 0.10 ± 0.022 0.06 ± 0.025 0.07 ± 0.023
ACC_rfLOO 0.59 ± 0.057 0.95 ± 0.032 0.94 ± 0.045 0.52 ± 0.062 0.99 ± 0.035 0.99 ± 0.036 0.99 ± 0.036
NMI_kmeans 0.035 ± 0.0085 0.035 ± 0.0089 0.035 ± 0.0093 0.0052 ± 0.0012 0.93 ± 0.054 0.92 ± 0.051 0.93 ± 0.055
Noise type
PCCintraclass 0.025 ± 0.0031 0.066 ± 0.0022 0.13 ± 0.014 0.013 ± 0.0029 0.32 ± 0.034 0.23 ± 0.030 0.31 ± 0.031
PCCinterclass −0.0040 ± 0.00097 −0.049 ± 0.0012 0.097 ± 0.0023 −0.011 ± 0.0058 −0.32 ± 0.031 −0.22 ± 0.029 −0.29 ± 0.033
PCCintraclass-PCCinterclass 0.029 ± 0.0032 0.12 ± 0.0024 0.033 ± 0.016 0.024 ± 0.0067 0.64 ± 0.061 0.45 ± 0.059 0.60 ± 0.063
Simintraclass 0.02 ± 0.0023 0.02 ± 0.0025 0.02 ± 0.0026 0.99 ± 0.0007 0.27 ± 0.012 0.25 ± 0.015 0.28 ± 0.013
Siminterclass 0 ± 0.0 0 ± 0.0 0 ± 0.0 0.99 ± 0.0014 0.16 ± 0.009 0.15 ± 0.011 0.16 ± 0.011
Simintraclass-Siminterclass 0.02 ± 0.0023 0.02 ± 0.0025 0.02 ± 0.0026 0.00021 ± 0.0020 0.11 ± 0.020 0.10 ± 0.026 0.12 ± 0.023
ACC_rfLOO 0.54 ± 0.023 0.98 ± 0.015 0.97 ± 0.021 0.93 ± 0.013 0.96 ± 0.015 0.95 ± 0.016 0.96 ± 0.018
NMI_kmeans 0.015 ± 0.0021 0.82 ± 0.0023 0.024 ± 0.0033 0.0042 ± 0.00056 0.83 ± 0.017 0.82 ± 0.018 0.82 ± 0.015
Noise size
PCCintraclass 0.023 ± 0.0031 0.048 ± 0.0015 0.026 ± 0.0033 0.0070 ± 0.00067 0.13 ± 0.038 0.09 ± 0.023 0.11 ± 0.031
PCCinterclass −0.0041 ± 0.00009 −0.028 ± 0.0012 −0.0071 ± 0.00013 −0.0045 ± 0.00021 −0.13 ± 0.037 −0.08 ± 0.025 −0.10 ± 0.033
PCCintraclass-PCCinterclass 0.027 ± 0.0031 0.076 ± 0.0024 0.033 ± 0.0034 0.012 ± 0.00069 0.26 ± 0.065 0.17 ± 0.049 0.21 ± 0.061
Simintraclass 0.02 ± 0.0014 0.02 ± 0.0011 0.02 ± 0.0015 0.99 ± 0.00002 0.26 ± 0.015 0.19 ± 0.016 0.23 ± 0.017
Siminterclass 0 ± 0.0 0 ± 0.0 0 ± 0.0 0.99 ± 0.00003 0.20 ± 0.017 0.16 ± 0.016 0.18 ± 0.018
Simintraclass-Siminterclass 0.02 ± 0.0014 0.02 ± 0.0011 0.02 ± 0.0015 0.00016 ± 0.0006 0.06 ± 0.027 0.03 ± 0.030 0.05 ± 0.033
ACC_rfLOO 0.59 ± 0.051 0.84 ± 0.034 0.82 ± 0.054 0.86 ± 0.041 0.91 ± 0.052 0.90 ± 0.053 0.91 ± 0.055
NMI_kmeans 0.024 ± 0.0081 0.58 ± 0.043 0.028 ± 0.0097 0.0019 ± 0.00091 0.56 ± 0.062 0.55 ± 0.055 0.57 ± 0.063
Partial clustering
PCCintraclass 0.049 ± 0.0038 0.046 ± 0.0042 0.06 ± 0.0021 0.027 ± 0.0023 0.13 ± 0.05 0.10 ± 0.04 0.12 ± 0.06
PCCinterclass 0.0010 ± 0.00085 −0.0011 ± 0.00092 −0.016 ± 0.0034 −0.025 ± 0.0026 −0.11 ± 0.023 −0.10 ± 0.024 −0.12 ± 0.025
PCCintraclass-PCCinterclass 0.048 ± 0.0043 0.048 ± 0.0047 0.079 ± 0.0070 0.053 ± 0.0049 0.24 ± 0.067 0.20 ± 0.062 0.24 ± 0.073
Simintraclass 0.02 ± 0.0041 0.02 ± 0.0044 0.02 ± 0.0063 0.99 ± 0.00011 0.27 ± 0.019 0.23 ± 0.020 0.25 ± 0.022
Siminterclass 0 ± 0.0 0 ± 0.0 0 ± 0.0 0.99 ± 0.00012 0.21 ± 0.021 0.18 ± 0.021 0.20 ± 0.021
Simintraclass-Siminterclass 0.02 ± 0.0041 0.02 ± 0.0044 0.02 ± 0.0063 0.00040 ± 0.00022 0.061 ± 0.033 0.052 ± 0.035 0.057 ± 0.038
ACC_rfLOO 0.86 ± 0.028 0.87 ± 0.026 0.93 ± 0.016 0.96 ± 0.029 0.90 ± 0.023 0.89 ± 0.031 0.91 ± 0.035
NMI_kmeans 0.65 ± 0.033 0.61 ± 0.032 0.90 ± 0.028 0.63 ± 0.035 0.57 ± 0.031 0.55 ± 0.035 0.59 ± 0.039
Conflicting clustering
PCCintraclass 0.095 ± 0.0052 0.095 ± 0.0051 0.095 ± 0.0061 0.032 ± 0.0063 0.62 ± 0.023 0.53 ± 0.033 0.59 ± 0.035
PCCinterclass −0.017 ± 0.0047 −0.020 ± 0.0049 −0.019 ± 0.0062 −0.0099 ± 0.0067 −0.20 ± 0.019 −0.18 ± 0.021 −0.19 ± 0.022
PCCintraclass-PCCinterclass 0.11 ± 0.011 0.12 ± 0.012 0.11 ± 0.013 0.042 ± 0.015 0.82 ± 0.037 0.71 ± 0.045 0.78 ± 0.051
Simintraclass 0.038 ± 0.0061 0.038 ± 0.0062 0.038 ± 0.0061 0.99 ± 0.00005 0.31 ± 0.024 0.27 ± 0.025 0.30 ± 0.029
Siminterclass 0 ± 0.0 0 ± 0.0 0 ± 0.0 0.99 ± 0.00007 0.11 ± 0.021 0.10 ± 0.020 0.12 ± 0.023
Simintraclass-Siminterclass 0.038 ± 0.0061 0.038 ± 0.0062 0.038 ± 0.0061 0.00019 ± 0.00010 0.20 ± 0.037 0.17 ± 0.039 0.18 ± 0.042
ACC_rfLOO 0.42 ± 0.020 0.51 ± 0.023 0.63 ± 0.031 0.93 ± 0.063 0.96 ± 0.034 0.94 ± 0.031 0.95 ± 0.036
NMI_kmeans 0.46 ± 0.033 0.49371 ± 0.034 0.84626 ± 0.045 0.11654 ± 0.081 0.92 ± 0.052 0.91 ± 0.053 0.92 ± 0.055
  1. The best performer was highlighted with the darkest color
  2. PCCintraclass: average Pearson correlation coefficients of patients within the same classes; PCCinterclass: average Pearson correlation coefficients of patients from different classes; Simintraclass: average similarity of patients within the same classes measured by the Gausian kernel;Siminterclass:average similarity of patients from different classes measured by the Gausian kernel;ACC_rfLOO: accuracy of leave-one-out cross-validation by random forest; NMI_kmeans: normalized mutual information between the true patient relationships and the clustering results by k-means