Extended Data Fig. 2: Technical overview.

a. Technical effect estimation. The data were regressed with the experimental variables (that is environmental parameters) and the technical ones (that is read length and number of reads). The plot shows the distribution of the betas resulting from the regression for the MG and MT ORF-based matrices. b. The three ORF-based omic quantification matrices are summarised by summing up the lines with the same ORF descriptor. The final result is a collection of 24 matrices + the original three. c. The six panels show the number of time-dependent EGs and the EG weights (equivalent to the Explained Variance) per omic in the nine summarisation matrices. The first EG (that is the basal state of the system) was removed and all the EG weights re-scaled per matrix. In the y axis ‘Fun’ stands for ‘Function’ and ‘Tax’ for ‘Taxonomy’. The number of selected EGs changes depending on the omic and the descriptor, however some trends can be seen in the EG weight. For MG and MT the EG weight is the largest, signifying that it is, if taken alone, the most informative layer of information. Interestingly in MT the second largest, with a decent margin, is the Species level, which can be explained as a level in which most of the individual genes information is conserved (that is genes of the same species will be expressed together over time). d. EG clustering. The columns represent the 17 EG clusters while rows indicate the different types of summarisation matrices. In the top panel the violin plots depict the distribution of the explained variance (EV) from the EGs in the cluster. The red dot indicates the maximal EV in the distribution and the EV of the cluster. On the y-axis there are the 27 matrices.