feat: use transposon-data to improve model

[sulheim]

Description of the issue:

By evaluating the false positive and false negative model predictions from the transposon-data evaluation it may be possible to improve the model further.

Current feature/value/output:

Note: This is just considering the transposon data

• FP: 31
• TN: 32
• TP: 92
• FN: 5

Identified problems and solutions:

• SCO5547 (currently FN). Can be solved by adding EC: 6.3.5.7 to the model

I hereby confirm that:

• This problem persists in the master branch of the repository
• A similar issue does not already exist