Open sam-israel opened 3 years ago
Actually, I see that it is definitely possible.
This package is devoted to analyzing high-throughput data (e.g. gene expression microarray, DNA methylation microarray, RNA-seq) from complex tissues.
My apologies. I understand that the microarray data should not be in log space?
In the vignette, it is not in log space
CBS_mix[1:4,1:4]
## X17.002 X17.006 X17.019 X17.023
## ABCB4 96.0 107.50 110.00 92.3
## ABCB9 98.3 109.75 103.85 92.1
## ACAP1 196.8 217.80 351.00 140.7
## ACHE 92.7 97.20 87.10 87.1
I have found the answer for it in the article
https://doi.org/10.1093/bioinformatics/btz196
Theoretically, mixing takes place at the raw scale and linear deconvolution is expected to work on the same scale. However, real biological data from high-throughput technology are complex and often include many sources of noise, distortion and anomaly that cannot be fully captured by simple parametric simulation. ... As a result, real data applications may find that analyzing the log-scale data delivers similar or even better performance
Is it possible to use TOAST for microarray data? Are there modifications or specifications that should be used when analyzing microarrays?