Neural network and kinetic modelling of human genome replication reveal replication origin locations and strengths

In human and other metazoans, the determinants of replication origin location and strength are still elusive. Origins are licensed in G1 phase and fired in S phase of the cell cycle, respectively. It is debated which of these two temporally separate steps determines origin efficiency. Experiments can independently profile mean replication timing (MRT) and replication fork directionality (RFD) genome-wide. Such profiles contain information on multiple origins’ properties and on fork speed. Due to possible origin inactivation by passive replication, however, observed and intrinsic origin efficiencies can markedly differ. Thus, there is a need for methods to infer intrinsic from observed origin efficiency, which is context-dependent. Here, we show that MRT and RFD data are highly consistent with each other but contain information at different spatial scales. Using neural networks, we infer an origin licensing landscape that, when inserted in an appropriate simulation framework, jointly predicts MRT and RFD data with unprecedented precision and underlies the importance of dispersive origin firing. We furthermore uncover an analytical formula that predicts intrinsic from observed origin efficiency combined with MRT data. Comparison of inferred intrinsic origin efficiencies with experimental profiles of licensed origins (ORC, MCM) and actual initiation events (Bubble-seq, SNS-seq, OK-seq, ORM) show that intrinsic origin efficiency is not solely determined by licensing efficiency. Thus, human replication origin efficiency is set at both the origin licensing and firing steps.Author summary: DNA replication is a vital process that produces two identical replicas of DNA from one DNA molecule, ensuring the faithful transmission of genetic information from mother to daughter cells. The synthesis of new DNA strands initiates at multiple sites, termed replication origins, propagates bidirectionally, and terminates by merging of converging strands. Replication initiation continues in unreplicated DNA but is blocked in replicated DNA. Experiments have only given partial information about origin usage. In this work we reveal the exact propensity of any site to initiate replication along human chromosomes. First, we simulate the DNA replication process using approximate origin information, predict the direction and time of replication at each point of the genome, and train a neural network to precisely recover from the predictions the starting origin information. Second, we apply this network to real replication time and direction data, extracting the replication initiation propensity landscape that exactly predicts them. We compare this landscape to independent origin usage data, benchmarking them, and to landscapes of protein factors that mark potential origins. We find that the local abundance of such factors is insufficient to predict replication initiation and we infer to which extent other chromosomal cues locally influence potential origin usage.