Results: 4

1.
Fig. 4.

Fig. 4. From: Kinannote, a computer program to identify and classify members of the eukaryotic protein kinase superfamily.

Performance of Kinannote and Kinomer HMMs on curated kinomes. Kinannote identification, partial classification and full classification results are shown in the second, third, and fourth columns, respectively. Kinomer identification and group-level classification are shown in the fifth and sixth columns, respectively. Curated kinase counts are shown in the first columns. For Kinannote and Kinomer, the columns are divided into TP, FN, FP categories; for the curated kinomes, the columns are divided into conserved, species-specific and unclassified categories. The colors associated with these categories are indicated in the key below the figure. The curated kinomes are from A, S.pombe; B, A.queenslandica; C, G.lamblia; and D, P.falciparum

Jonathan M. Goldberg, et al. Bioinformatics. 2013 October 1;29(19):2387-2394.
2.
Fig. 1.

Fig. 1. From: Kinannote, a computer program to identify and classify members of the eukaryotic protein kinase superfamily.

Algorithm used by Kinannote to produce a draft kinome from a predicted proteome. Computational steps are numbered and sequence sets are indicated by letters. The aggregate of sets comprising the draft kinome are indicated by the shaded area. Unclassified kinases may be species-specific or members of novel families; twilight hits are recorded but are not part of the draft kinome. ePKs, eukaryotic protein kinase superfamily members; aPKs, atypical protein kinases

Jonathan M. Goldberg, et al. Bioinformatics. 2013 October 1;29(19):2387-2394.
3.
Fig. 2.

Fig. 2. From: Kinannote, a computer program to identify and classify members of the eukaryotic protein kinase superfamily.

Receiver-operator characteristic (ROC) curves for identification of protein kinases in reference kinomes by Kinannote. The cutoff score (HMM score for kinase identification, Table 1) of results from a search of the comprehensive predicted gene set against a protein kinase hidden Markov model (Fig. 1.1, cutoff applied at Fig. 1.5) was varied from stringent (left side of range) to permissive (right side of range). Blue, S.pombe; orange, A.queenslandica; green, C. G.lamblia; purple, P.falciparum. The upper left corner of the plot is expanded in the inset. Points representing the optimum setting of −66 are circled. TPR, true positive rate; FDR, false-discovery rate

Jonathan M. Goldberg, et al. Bioinformatics. 2013 October 1;29(19):2387-2394.
4.
Fig. 3.

Fig. 3. From: Kinannote, a computer program to identify and classify members of the eukaryotic protein kinase superfamily.

ROC analysis to identify optimum settings for classification by Kinannote and compare classification methods. Circles: effect on classification of varying the number of consistent consecutive top BLAST hits of candidate kinases against the reference database. The BLAST search is described in Figure 1.3, and the criterion is applied at Figure 1.7. The number of consistent hits required for classification ranged from 5 (left side of range) to 1 (right side of range). Partial classification is indicated by half-filled symbols, and full classification is indicated by filled symbols. The points obtained at three consistent consecutive hits are indicated with boxes. This setting was chosen as the default value for Kinannote because it provides a good compromise between sensitivity and precision. Diamonds: effect on classification of varying the E-value cutoff from the BLAST search (Fig. 1.6) from 1 × 10−100 (left side of range) to 1 (right side of range). As aforementioned, partial and full classification are indicated by half-filled and filled symbols, respectively. The test genomes are A, S.pombe; B, A.queenslandica; C, G.lamblia; and D, P.falciparum. TPR, true-positive rate; FDR, false-discovery rate

Jonathan M. Goldberg, et al. Bioinformatics. 2013 October 1;29(19):2387-2394.

Supplemental Content

Recent activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...
Write to the Help Desk