Regarding Cij matrix calculatd by LMI approach
Issue #393
resolved
Dear sir,
i have calculated Cij matrix both by dccm(pearsson like) method and LMI method.. In dccm method i got _ve correlatons (ranges from -1:1) but in LMI calculations all the values are in the range 0:1(ie. no -ve correlations). so can you please physically simplify this.
And also which one is more appropriate?
Thanks
Comments (2)
-
-
- changed status to resolved
This is question not a bug.
- Log in to comment
Hi,
Those ranges of values are the nature of the two correlation methods. For Pearson, both 'parallel' and 'anti-parallel' correlations are allowed, and so the range is from -1 to 1. For LMI, there is no direction of correlation, and so the range 0-1 indicates the magnitude only. You can find more details about LMI here (http://onlinelibrary.wiley.com/doi/10.1002/prot.20784/abstract)
There is no clear answer which one is more appropriate. To our experience, usually LMI will give an overall stronger correlation but the relative strength (or the ranking) is similar to Pearson. You could try both methods and see which one fits better for your case. Or, just choose Pearson because it is simpler.