how to run PCA analysis for genetic data (1 for present, 0 for absent alles) with NTSYS please
hi, dear Nabila El Kadri
follows steps: 1. Compute correlations among variables Similarity - simint - cofficent= corr - "Output the correlation matrix"
2. Extract first 3 PCA axes from correlation matrix Ordination - eigen Output principal component axes
3. Project objects onto PCA axes Ordination - Project
Output projections is PCA
PCA Instructions on NTSYS
The steps:
- Realize the rectangular matrix with values in the absence of commas first in the pad and then acquire it in the NtEdit
- Standardize values with different units with the "stand" function
- With the "simint" function, a correlation matrix is obtained between OTUs
- With the "eigen" function, PCAs are obtained
- With the "Proj" function you get the graphical representation of PCAs
The alternative is to run a batch file obtained by copying instructions from the NTSYS guide to the "PCA" entry by calling the test file with the .ntb extension (or batch file) and saving it to the same folder as the file NTedit obtained from the rectangular array.
WARNING: This procedure provides the eigenvectors file (weight of variables on the individual PCA axes) but does not return the eigenvalues (ie the values that explain the variability described by each variable).
To achieve the eigenvalues, one should step in:
- open NTSYS after running the batch file
- go to the eigen function and load the file with the correlation matrix (normally called corr.nts) previously obtained by the batch file. The function "promises" to spit out eigenvalues in a file that will be white.
- Therefore, at the end of the processing, since the eigenvalue file will be white, it will be necessary to copy the text screen that NTSYS returns at the end of the processing itself and where the values with the corresponding variance and cumulative variance are reported!
- Badges
- Science topic