05. We analysed these physico-chemical variables of the pitviper venom PLA2s by DFA in SPSS v.14, using functional activities as groups and individual PLA2 toxins as cases. Data on functional activity were primarily gathered
from UniProt. However, it has previously been noted that many database protein entries are not annotated with function Venetoclax research buy ( Tan et al., 2003), there are no actively maintained databases specifically for snake venom toxins, and the only current database on animal toxins has limited functionality ( Jungo et al., 2012). Therefore, we also carried out searches of the primary literature using GoPubMed (www.gopubmed.org). Reported functional activities of PLA2s are very varied; 15 are listed by Kini
(1997) while Doley et al. (2009), mention at least 12 distinct activities. For simplicity, we reduced the number of activities to the six most commonly reported, i.e., neurotoxic, myotoxic, antiplatelet, anticoagulant, oedematous, and hypotensive. Variables were entered together and posterior probabilities of group membership (including for the ungrouped proteins, which did not take part in the discrimination, but whose position relative to the calculated axes was also calculated) were saved. A sequence profile represents the information contained in a multiple sequence alignment as a table of position-specific symbol comparison values and gap penalties. The profile-based neighbour-joining (PNJ) method Dabrafenib is a means of obtaining more resolution in a large tree by successively collapsing clusters supported above a certain user-determined value into a summary profile. It is claimed to be as accurate as Bayesian methods, but much more computationally efficient (Müller et al., 2004). We used ProfDistS v0.9.8 (Wolf et al., 2008), with general time-reversible distances based on the VTML model, which models protein evolution as a Markov process (Müller and Vingron, 2000). Profiles were built for clusters with either sequence
identity above 97% or bootstrap values PJ34 HCl (from 500 bootstraps of the initial NJ tree) of greater than 70% in an iterative process (Merget and Wolf, 2010). The resulting PNJ tree was rooted and annotated in Dendroscope 3 (Huson and Scornavacca, 2012). It is important to note that the resulting tree reflects the degree of structural similarity among amino-acid sequences, and will not necessarily reflect evolutionary relationships among the sequences (i.e., it is not a gene tree) since the non-coding parts of the gene may be quite divergent. A multitude of computational tools are available for the prediction of molecular function based on de novo protein sequences ( Punta and Ofran, 2008). The more powerful programs combine several approaches. One of these, Protfun (available at http://www.cbs.dtu.dk/services/ProtFun-2.