Fig. 2: Transcriptional profiling representation learning by TranSiGen. | Nature Communications

Fig. 2: Transcriptional profiling representation learning by TranSiGen.

From: Deep representation learning of chemical-induced transcriptional profile for phenotype-based drug discovery

Fig. 2

a Performance of TranSiGen in transcriptional profiling reconstruction and prediction (n = 15,713). Box-and-whisker plots show the median (center line), 25th, and 75th percentile (lower and upper boundary), with 1.5 × inter-quartile range indicated by whiskers and outliers shown as individual data points. b The change of TranSiGen’s performance for \(\varDelta X^{\prime}\) with the correlation between \({X}_{1}\) and \({X}_{2}\). Box-and-whisker plots show the median (center line), 25th, and 75th percentile (lower and upper boundary), with 1.5 × inter-quartile range indicated by whiskers and outliers shown as individual data points. The line plot corresponds to the sample size within each threshold, and the specific sample size is shown in source data. c Dimensionality reduction visualization using \(\varDelta X\) and TranSiGen-derived \(\varDelta X^{\prime}\) for different cell lines. d Distribution of Pearson’s correlation coefficients of profiles for the same MOA by \(\varDelta X\) and TranSiGen-derived \(\varDelta X^{\prime}\). Source data are provided as a Source Data file.

Back to article page