Average run time and memory usage for inserting/placing 50 samples into the 16S 800-taxa reference tree.

Tests were run on a server with 64 Intel Xeon Gold 6242 CPU cores and 1500 GB RAM for 10 repeated runs. The performance of TIPars and UShER were also compared on a general computer with 8 Intel Xeon Gold 6242 CPU cores. TIPars ran with JAVA setting -Xmx1G loaded with FASTA files. “Insertion time” indicates the time for the key step of searching for the best position to place or insert only a query sequence, while “total runtime” is the total time to run the program. EPA-ng insertion time data is not presented as its reporting of this time is limited to whole seconds and given as 0 in this case. MAPLE is a single-thread program indicating by 1 CPU core assigned. N/A indicates that data are not applicable. Total runtimes for the different programs consist of 1) TIPars: “insertion time” (IT) + I/O processing (I/O); 2) UShER: IT + I/O + ‘mutation-annotated tree’ computation; 3) EPA-ng: IT + I/O + likelihood weight ratio (LWR) computation; 4) RAPPAS: IT + I/O, 5) APPLES-2, IQTREE2 and PAGAN2: their total run times and 6) MAPLE: IT+I/O+‘substitution matrix’ (without topology refinement). * The run time for RAPPAS to compute its ‘pkDB’ database (3095.31 seconds on average) is not included in “Mean total run time”.
(XLSX)