Fig. 4: Applying our DLM to the prediction of single-plex hybridization and strand displacement rate constants.
From: A deep learning model for predicting next-generation sequencing depth from DNA sequence
Schematic of (a) hybridization and (b) strand displacement reactions. c Sample kinetics traces of hybridization and a strand displacement reaction. Reaction yields inferred through fluorescence; see Supplementary Note 2 for experimental and data processing details. d Accuracy of DLM predictions of hybridization (blue) and strand displacement (orange) rate constants. Dark gray shading marks the zones where the predicted and the observed read depth agreed to within a factor of 2; light gray shows agreement to within a factor of 3. Due to the small number of data points here (421), we implemented prediction based on 100-fold leave-one-class-out (LOCO) rather than 20-fold cross-validation, due to the large expected variation in small validation classes. Note that strand displacement and hybridization reaction parameters were co-trained using the same DLM, and predictions were likewise made using a single DLM. e Comparing DLM prediction performance to a previous expert system machine learning approach based on weighted neighbor voting8. In box-whisker plots, the central mark indicates the median, and the bottom and top edges of the box indicate the 25th and 75th percentiles, respectively. The maximum whisker lengths are specified as 1.5 times the interquartile range. Prediction results are similar.
