The results page for Foldseek reports two different TM-scores for the same target-query entry. One is display in the table and the other is displayed in the pop out alignment viewer (see attached png). For the structures I am working with, the alignment view TM-score is almost always higher than the one displayed in the table.
TM-scores vary depending on which length was used to normalize to (query, target, average) so that's my guess of where the contradiction comes from. Either way, having the scores agree and indicating how the TM-score was normalized would be nice.
Also, when exporting the Hit table for the result, the m8 file agrees with the TM-score in the table.
Example 1
Example2
The results page for Foldseek reports two different TM-scores for the same target-query entry. One is display in the table and the other is displayed in the pop out alignment viewer (see attached png). For the structures I am working with, the alignment view TM-score is almost always higher than the one displayed in the table.
TM-scores vary depending on which length was used to normalize to (query, target, average) so that's my guess of where the contradiction comes from. Either way, having the scores agree and indicating how the TM-score was normalized would be nice.
Also, when exporting the Hit table for the result, the m8 file agrees with the TM-score in the table.
Example 1
Example2