Confirmatory Factor Analysis (CFA) to reduce the large number of variables into a few. Then Structural Equation Modelling (SEM) may be good.

There is an important difference between individual, Likert-scored items and multi-item Likert scales. If you are using multi-item scales, you either need to add together the items or use a Structural Equation model.

In terms of the different scoring you have for the various variables, this doesn't have any practical effect on the analysis itself. In particular, it will affect how you interpret "metric" regression coefficients, but will have no effect on correlations or standardized regression coefficients.

https://www.researchgate.net/post/How_can_I_convert_different_point_Likert_scales_for_all_questionnaires_in_a_survey_so_that_they_are_the_same

Likert scales having odd choices: positive, negative, and a neutral, or undecided can easily be transformed but the Likert scale with even choices is very tricky as such scale forces the respondent to express his/her choice either way (positive or Negative) and can yield groupings. For applying statistics, standardization is must. Few recommend calculating z-scores whereas few recommends to apply CFA

Following link will be helpful

http://www-01.ibm.com/support/docview.wss?uid=swg21482329

Thank you everyone for their guidance. It really help me alot.

If items of constructs are adopted from previous validated studies then use likert scale of each construct "as it is with no transformation in likert scales".

Thank you