DNA is a complex molecule carrying the instructions an organism needs to develop, live and reproduce. In 1953, Watson and Crick discovered that DNA is composed of two chains forming a double-helix. Later on, other structures of DNA were discovered and shown to play important roles in the cell, in particular G-quadruplex (G4). Following genome sequencing, several bioinformatic algorithms were developed to map G4s in vitro based on a canonical sequence motif, G-richness and G-skewness or alternatively sequence features including k-mers, and more recently deep learning. Here, we propose a novel convolutional neural network (DeepG4) to map active G4s (forming both in vitro and in vivo). DeepG4 is very accurate to predict active G4s, while most state-of-the-art algorithms fail. Moreover, DeepG4 identifies key DNA motifs that are predictive of G4 activity. We found that active G4 motifs do not follow a very flexible sequence pattern as current algorithms seek for. Instead, active G4s are determined by numerous specific motifs. Moreover, among those motifs, we identified known transcription factors which could play important roles in G4 activity by contributing either directly to G4 structures themselves or indirectly by participating in G4 formation in the vicinity. Lastly, variant analysis suggests that SNPs altering predicted G4 activity could affect transcription and chromatin, e.g. gene expression, H3K4me3 mark and DNA methylation. Thus, DeepG4 paves the way for future studies assessing the impact of known disease-associated variants on DNA secondary structure by providing a mechanistic interpretation of SNP impact on transcription and chromatin.