possible_issues.md

Potential Issues

This list contains potential bugs & problems detected in the original code while refactoring.

• check_element detects C if formula contains Cl -> No compounds are filtered because they don't contain carbon
• multilevelannotationstep5:
  • the .combine = rbind parameter of foreach::foreach() function causes a twisted output, collapsing a 174x60 matrix to a 61x10 matrix -> 61 are the non-empty rows, but the entries in rows that have more than 10 columns filled get discarded;
  • the issue described above may cause multilevelannotationstep5 to compute different outputs depending on whether the loop is executed in parallel (%dopar%) or sequentially (%do%). The results may differ in chemical scores or the number of annotations. When chemical scores differ, sequential execution will systematically output lower scores.
• in multilevelannotationstep4 (and functions called by it) are some suspicious conditions where they check for curdata$score < 10 and curdata$score > 10, but never for actual value 10
• in get_chemscorev1.6.71 in the isotope detection, the RT clustering is recomputed with a fixed RT diff of 10 seconds, changing the whole previous RT clustering.
• in get_chemscorev1.6.71 this part for the calculation is never called under current tests.
• In get_chemscorev1.6.71 the correlation matrix is apparently only read on the diagonal, which implies that values will always be 1. The function could also just return a matrix of 1s with the size of mzid_cur x mzid_cur