Skip to main content
Fig. 4 | BMC Bioinformatics

Fig. 4

From: Improving prediction of phenotypic drug response on cancer cell lines using deep convolutional network

Fig. 4

The performance with different percentages of data used. The x-axis is the percentage of data used as training data from the total existing IC50 values (172114) in the database. Since there is 10% for validating and 10% for testing, the max x is 80%. The y-axis is the top-quartile performance of our model. The solid lines represent the result on total existing data, and the dash lines represent the results where only the IC50 values below the max screening concentration threshold(max_conc) are used, below which the data is more accurate. Since there are only 64,440 values below max_conc, so the dash lines end at around \(\frac {64,440}{172,114}*80\% = 30\%\)

Back to article page