1) Run all preprocessing steps as usual.

2) Run cora best-mapping search pipeline with the desired command-line parameters, but define '-C 1110' that excludes the mapping inference (final) stage.

3) Once finished, run the same command with '-C 0001', but exit prematurely by killing the process (e.g. Ctrl+C)

4) Copy the command printed on screen that starts "mappingInference" executable and ends with two identical integers (e.g. 2 2). The same command is also listed in OUTPUT_run_log_file.search log file.

5) If you want to perform best-mapping with maximum distance of 12 (for each mate), modify the "mappingInference" executable name to "mappingInference_E6" and change the last two integers to "6 6". Similarly if you want to perform best-mapping with a maximum distance of 10, then modify the executable name to "mappingInference_E5" and change the last two integers to "5 5". For distance 8, do it similary with 4s instead.

6) Run the modified command.

The output should contain the higher error distance mappings as long as one of the k-mers in the read had a match to the reference within the similarity threshold of the original call.

This mod is tested to be working on BEST_FAST mapping mode for Hamming/Levenshtein distances (with almost no increase in total runtime, but possibly using more memory). For Levenshtein distance, this mod also works with BEST mode, but possibly with some noticable increase in runtime. For any other mapping mode (including ALL and BEST_SENSITIVE modes), unpredictable behavior could occur.