Problem:"RAMClustR with csv works" test fails on M1 Mac due to different (from Intel chips) floating numbers handling when computing mz_rt features. The difference lies in rounding error; however, no difference is tolerated because the features are stored as strings in the form mz_rt.
Workaround:mz_rt labels are now checked separately from the whole dataset. Each feature label is disassembled into individual mz and rt, these are then converted to numeric datatypes and checked against their expected values.
Problem: "RAMClustR with csv works" test fails on M1 Mac due to different (from Intel chips) floating numbers handling when computing mz_rt features. The difference lies in rounding error; however, no difference is tolerated because the features are stored as strings in the form mz_rt.
Workaround: mz_rt labels are now checked separately from the whole dataset. Each feature label is disassembled into individual mz and rt, these are then converted to numeric datatypes and checked against their expected values.