morphmet
Tue, 17 Nov 2009 09:17:23 -0800
-------- Original Message -------- Subject: RE: MorphoJ: 2B-PLS vs Modularity hypothesis Date: Tue, 17 Nov 2009 05:42:37 GMT From: Javier Santos <javiersant...@hotmail.com> To: <morphmet@morphometrics.org> References: <4af521e2.1070...@morphometrics.org> Hello again, After reading the papers mentioned before I still have a few doubts. Correct me when wrong: If I want to test for modularity I use the MorphoJ modularity hypothesis to search for hypothetical modules. Modules should have a low RV value --> which means that there is a low degree of covariation between the LM configurations that constitute modules. Now, if I run a Two-block PLS in MorphoJ between a pair of hypothetical modules I also obtain a RV coefficient. This coefficient gives the overall correlation between the covariance matrixes of the two blocks. It seems to me that this RV coefficient in 2B-PLS also indicates the degree of covariation between modules, from here that I find it confusing how to interpret this RV coefficient: Should I focus on low RV values in 2B-PLS (= low degree of covariation between blocks) to establish hypothetical modules like I do in the modularity hypothesis? However, from the earlier mail I understood that the RV coefficient in 2B-PLS is different (maybe in calculation?) from that of the modularity hypothesis (when only using 2 hypothetical modules), is that correct? If indeed this is the case, then would a high RV coefficient in 2B-PLS indicate that the two blocks under study may constitute independent modules? Why do you say PLS does not permit to search for modularity if it also examines the covariation between sets of shape variables as does the modularity hypothesis? I am quite confused, any help is appreciated. Many thanks in advance. All the best, Javier Santos jssan...@usal.es <mailto:jssan...@usal.es> <mailto:jssan...@usal.es> University of Salamanca > Date: Sat, 7 Nov 2009 02:29:38 -0500 > From: morphmet_modera...@morphometrics.org > To: morphmet@morphometrics.org > Subject: Re: MorphoJ: PLS vs Modularity hypothesis > > > > -------- Original Message -------- > Subject: Re: MorphoJ: PLS vs Modularity hypothesis > Date: Thu, 05 Nov 2009 11:02:11 +0100 > From: carlo.mel...@unina.it > To: morphmet <morphmet_modera...@morphometrics.org> > References: <4aef15fb.5070...@morphometrics.org> > > > > Dear Javier, > > PLS does not allow you to test for modularity. It will maximise the > covariation between two block of variables that you can assume they > could be modules. But a strong correlation does not mean that they are > modules. The PLS will extract pairs of SA (which are vectors: Singular > Axis) that are associated with an SV (Singular Value) extracted from a > Singular Value Decomposition (basically the Singular Values describe the > amount of covariance explained by each pair of SA). > > I strongly suggest you to read: "/Rohlf F.J., Corti M. 2000. Use of > two-block partial least squares to study covariation in shape. > Systematic Biology 49: 740?753." where the tecnique is explained much > better than I did./ > > Probably you want to test for modularity through Morphoj and the > Escoufier index (RV, Escoufier 1973) that is different metric not > entirely correlated with PLS. Check > > Klingenberg, C. P. 2009. Morphometric integration and modularity in > configurations of landmarks: Tools for evaluating a-priori hypotheses. > Evolution & Development 11:405?421. > > where you can found more info on how to test for modularity. > > Good luck and if you need the pdf just ask. The Klingenbergh pdf is > linked in MorphoJ help manual. > > All the best > > Carlo Meloro > > > > Quoting morphmet <morphmet_modera...@morphometrics.org>: > > > > > > > -------- Original Message -------- > > Subject: MorphoJ: PLS vs Modularity hypothesis > > Date: Sun, 1 Nov 2009 06:13:43 -0800 (PST) > > From: Javier Santos <javiersant...@hotmail.com> > > To: <morphmet@morphometrics.org> > > > > > > > > Hello everybody, > >> > I had a doubt considering how to correctly interpret the RV coefficient
> > given in MorphoJ when doing PLS and Modularity hypothesis analysis.> > Should the RV coefficient given in PLS analysis be interpreted the same
> > as the RV coefficient in the Modularity hypothesis? So, should low RV > > coefficients given in PLS analysis with significant p-values be > > considered as support that blocks constitute seperate modules? Also, can > > someone give any more info on how these two analysis differ when > > studying morphological integration and modularity, apart from the fact > > that in PLS modules are chosen a priori. Thank you. > > > > > > All the best, > > Javier Santos > > jssan...@usal.es <mailto:jssan...@usal.es> > > University of Salamanca > > > > ------------------------------------------------------------------------> > Windows Live: Friends get your Flickr, Yelp, and Digg updates when they
> > e-mail you. > > > <http://www.microsoft.com/middleeast/windows/windowslive/see-it-in-action/social-network-basics.aspx?ocid=PID23461::T:WLMTAGL:ON:WL:en-xm:SI_SB_3:092010> > > > > -- > > Replies will be sent to the list. > > For more information visit http://www.morphometrics.org > <http://www.morphometrics.org/> > > > -- > Replies will be sent to the list. > For more information visit http://www.morphometrics.org > ------------------------------------------------------------------------ Keep your friends updated— even when you’re not signed in. <http://www.microsoft.com/middleeast/windows/windowslive/see-it-in-action/social-network-basics.aspx?ocid=PID23461::T:WLMTAGL:ON:WL:en-xm:SI_SB_5:092010> -- Replies will be sent to the list. For more information visit http://www.morphometrics.org