Graul RC and Sadee W Evolutionary Relationships Among G Protein-Coupled Receptors Using a Clustered Database Approach AAPS PharmSci 2001; 3 (2) article 12 (https://www.pharmsci.org/scientificjournals/pharmsci/journal/01_12.html).

Figures and Tables

Table 1. INCA sequence clustering. (Click here to view)

Table 2. BLAST sub clustering (Click here to view)

Table 3. BLAST results. (Click here to view)

Table 4. PSI-BLAST results. (Click here to view)

Table 5. IMPALA results. (Click here to view)

Table 6. Standard Pfam/HMMER results. (Click here to view)

Figure 1.Error vs. coverage for BLAST, IMPALA, and Pfam/HMMER. We compared the performance of BLAST, IMPALA, and Pfam/HMMER. We used the 5h1a and vipr clusters as well as the corresponding PPSMs and Pfam models to analyze our database sequences. We considered the 1285 sequences from the following 19 clusters as GPCRs, or true positives: 5h1a, 5ht, car1, friz, latr, oa1, odr, odr10, olf1, pe22, sra1, srb6, srd1, sre1, srg1, sro1, ste3, vipr, and vn1. We considered the 192 sequences from the following 4 clusters as non-GPCRs, or false positives: er21, patc, pet1, and psn1. Each of these 1477 sequences was used as a query against both the 5h1a (168 sequences) and vipr (142 sequences) clusters. We collected selected E-values. For BLAST, the top sequence hit was used. For IMPALA, the top scoring PPSM hit was used. For HMMER/Pfam, a single profile HMM hit was used. Error rate is plotted against coverage (fraction of true positives). Hits were sorted in increasing order of E-value. Data points represent the tradeoffs between fraction of true positives and fraction of false positives at different E-value cutoffs. Generally, BLAST outperformed the other methods over the range of E-values. Figure 1A. For Pfam, the 7tm_1 model was used. For IMPALA, our set of 5h1a PPSMs was used. Generally, our 5h1a PPSMs outperformed the Pfam 7tm_1 models over the range of E-values. Figure 1B. For Pfam, the 7tm_2 model was used. For IMPALA, our set of vipr PPSMs was used. Generally, our vipr PPSMs outperformed the Pfam 7tm_2 hmmfs (fragment) model over the range of E-values; however, the Pfam 7tm_2 hmmls (global) model outperformed our vipr PPSMs.

Figure 2.Cluster dendrogram. Data from BLAST (Table 3), PSI-BLAST (Table 4) and IMPALA (Table 5) were used to compute a cluster dendrogram. Minimum E-values were stratified according to the series: 0, 1e-128, 1e-64, 1e-32, 1e-16, 1e-8, 1e-4, 1e-2, 1e-1, 1e0, mapped to values from 0 to 9, 0 representing closest clusters and 9 representing distant clusters, and consolidated into a single matrix. The resultant matrix was analyzed using the Fitch program from the PHYLIP (https://evolution.genetics.washington.edu/phylip.html ) software package. All branches were set to uniform length, so as not to imply that the dendrogram represents actual evolutionary distance. The dendrogram was subsequently drawn using the Drawtree program, also from the PHYLIP software package. Clades are annotated (colored) with the highest organism represented within them.