| Abstract [eng] |
This work explores possible ways to improve the quality of the ColabFold program results. One of them is the extension of new protein sequence search algorithms. Current ColabFold solution only allows sequence search using MMSeq2 algorithm. A solution is presented that allows to easily add specified algorithms and thus extend the ColabFold application. In this work ColabFold has been extended with the DIAMOND and SWIPE algorithms. Another proposal is to combine the results of MMSeq2 with one of the other protein search algorithms. The added results can be filtered using a sequence compatibility index, which is computed with an ensemble structure of machine learning models. |
