compareMS2 tree drawing fails (for large trees?)

[magnuspalmblad]

I ran compareMS2 on 526 (fairly large - MGFs around 250 MB) datasets (138,075 comparisons). All comparisons seems to have been made, and the distance matrix (.meg) file opens in MEGA X (10.2), displaying all 526 datasets in the tree. Perhaps the error reported by compareMS2 has to do with the large number of datasets being compared? Or the size of the matrix? The last row has 525 columns, 4861 characters in the .meg file.

This is the error (or errors) as displayed in the GUI:

[524D]

That doesn't look good!

File compareMS2_to_distance_matrices.c contains the lines:

#define MAX_SAMPLES 120
#define MAX_SPECIES 120
#define MAX_COMP_BETWEEN_SPECIES 120

So, it's surprising that it didn't fail much earlier. These constants are used in memory allocation, and I can only guess that it worked because memory was allocated abundantly. In any case, I've increased the number to 4096, which is means 16M comparisons, that should be sufficient for all cases.

I'll produce a new intermediate release later today.

[magnuspalmblad]

Should we allocate the memory dynamically - like we do for the spectra? I.e. compareMS2_to_distance_matrices.c should first read the list of files, count how many samples (and species) there are, and then allocate the memory that is needed?

[524D]

The new version is now at version 2.2.0-beta.2. I hope the issue is solved for now.

Indeed it would be better to so solve this properly, but there are still quite a bit of fixed size items in the code. I think it makes sense to rewrite the code in a modern programming language where memory management is simpler.