A function to replace NA in the data frame using
metabolomic Non-negative Matrix Factorization.
Arguments
- missdf
a data frame with missing values to be imputed containing features in columns and samples in rows.
- kgroup
the range of k value.
- initialType
type of pre-imputation. Possible values are:
mean,median, andzero.
k
If k is not defined, it becomes a range between 1 and the minimum of
number of columns and number of rows of missdf as advised in the
original article (see References).
References
Xu J, Wang Y, Xu X, Cheng K, Raftery D, Dong J (2021). “NMF-Based Approach for Missing Values Imputation of Mass Spectrometry Metabolomics Data.” Molecules, 26(19), 5787. ISSN 1420-3049, doi:10.3390/molecules26195787 .
Examples
data(sim_miss)
impute_mnmf(sim_miss + 100)
#> X1 X2 X3 X4 X5 X6
#> 1 100.1551 100.8812 100.4637 100.1195 100.6004 100.5534
#> 2 100.9684 100.6075 100.6338 100.6181 100.0081 100.7521
#> 3 100.4683 100.5745 100.2476 100.2983 100.3701 100.5832
#> 4 100.7768 100.8031 100.5513 100.1460 100.7242 100.8478
#> 5 100.4079 100.7991 100.2348 100.1897 100.4183 100.7087
#> 6 100.5388 100.8019 100.2586 100.5755 100.2382 100.4700
#> 7 100.8301 100.7521 100.9529 100.0002 100.8879 100.4883
#> 8 100.1871 100.5755 100.8569 100.3283 100.5778 100.3472
#> 9 100.7800 100.9442 100.6595 100.0783 100.7400 100.2030
#> 10 100.1939 100.2190 100.5545 100.6042 100.1607 100.7400
#> 11 100.4342 100.4780 100.8769 100.3268 100.5296 100.8937
#> 12 100.0023 100.0447 100.8019 100.7905 100.5745 100.5745
#> 13 100.8347 100.1685 100.5467 100.8301 100.9601 100.0844
#> 14 100.7042 100.3683 100.2328 100.8101 100.3986 100.1483
#> 15 100.9570 100.5633 100.6023 100.2028 100.4085 100.3703
#> 16 100.9485 100.7202 100.7125 100.5060 100.6133 100.8612
#> 17 100.6007 100.7343 100.6489 100.5207 100.7066 100.8367
#> 18 100.2618 100.7037 100.1876 100.0843 100.7991 100.7405
#> 19 100.6430 100.7368 100.2384 100.2688 100.1613 100.9858
#> 20 100.5262 100.5922 100.7671 100.8147 100.8859 100.0321