Add semsim-comparison percentage-diff metadata columns

[souzadevinicius]

The current semsim-comparison percentage-diff file encodes metadata file as column name like this:

subject_id	object_id	phenio-hp-hp.semsimian.0.7_jaccard_similarity	upheno1-hp-hp.semsimian.0.7_jaccard_similarity	diff
HP:0010136	HP:0010147	0.8021978021978022	0.8823529411764706	9.99%
HP:0010147	HP:0010136	0.8021978021978022	0.8823529411764706	9.99%
HP:0008060	HP:0007750	0.8524590163934426	0.9375	9.98%
HP:0007750	HP:0008060	0.8524590163934426	0.9375	9.98%

It needs to be replaced to this:

subject_id	object_id	jaccard_similarity_left	jaccard_similarity_right	diff	left_file	right_file
HP:0010136	HP:0010147	0.8021978021978022	0.8823529411764706	9.99%	phenio-hp-hp.semsimian.0.7	upheno1-hp-hp.semsimian.0.7
HP:0010147	HP:0010136	0.8021978021978022	0.8823529411764706	9.99%	phenio-hp-hp.semsimian.0.7	upheno1-hp-hp.semsimian.0.7
HP:0008060	HP:0007750	0.8524590163934426	0.9375	9.98%	phenio-hp-hp.semsimian.0.7	upheno1-hp-hp.semsimian.0.7
HP:0007750	HP:0008060	0.8524590163934426	0.9375	9.98%	phenio-hp-hp.semsimian.0.7	upheno1-hp-hp.semsimian.0.7

PS: Some stats

A file with 265829 lines had its size increased from 16MB using the first template to 30MB compared to the second one.

[matentzn]

Thank you @souzadevinicius I am not sure what is the right answer. I reached out for other opinions. Maybe your original solution (solution 1) is actually preferable after all if the file size increase is so enormous. On the other hand, maybe adding a column compresses well? These 16MB and 30MB files, if you gzip them, what is their difference?

[souzadevinicius]

Replaced by this approach
monarch-initiative/monarch-semantic-similarity-profiles#9