·   Wiki Home
 ·   Data Processing
 ·   Hemileia vastatrix
 ·   Hypothenemus hampei
 ·   Coffea
 ·   Beauveria bassiana
 ·   Title List
 ·   Uncategorized Pages
 ·   Random Page
 ·   Recent Changes
 ·   Wiki Help
 ·   What Links Here

Active Members:



Create or Find Page:


View Secretome Proteins


As part of parasitic colonization, different plant pathogens, including bacteria, oomycetes, fungi and nematodes, secrete effector proteins which are transported inside host cells. The evidence about the relation between effector proteins and the modulation of the plant defense circuitry sheds light on molecular fundations of infection-response process of plant-pathogen interaction and therefore, provides an important way to analyze data result of whole-genome sequencing. An ab initio identification of putative secreted effectors through comparative secretome analysis and Blast searches is limited because of a general low homology between organisms. Althought sufficient homology has been reported for some cases, as RXLR motif in Phytophtora species (Win et al., 2007); Avr4, Ecp6 and the functional homologue of ToxA peptide from S. nodorum in P. tritici-repentis (Friesen et al., 2006) or MISSPs effector in Laccaria bicolor (Martin et al., 2008), a deep knowledge about the structure, diversity and conservative regions in effector proteins still in development.
The methodology used to identify in silico candidate secretome genes for the genome of Hemileia vastatrix followed essentially a three step procedure:
1. Gene prediction using AUGUSTUS predictor.
2. Search for secretome patterns using the statistical models of SignalP 3.0 and TargetP 1.1 softwares.
3. Annotation to predicted gene structures based on homology to previously annotated genes.

Gene Prediction


Hidden Markov models form the basis for most gene-prediction algorithms. From: How does eukaryotic gene prediction work? Michael R Brent. Nature Biotechnology 25, 883 – 885 (2007)

Augustus predictor
Like others gene predictors, utilizes a statistical model of gene structure that require training on each organism for accurate prediction. In this initial phase of coffee rust project, we used the program already trained on other fungi sequences. We included 33599 contigs from Newbler Assembler (length > 210 bp) and ran Augustus predictor in both strands and partial gene options. A total of 2888 sequences predicted that encode 70 o more amino acids were obtained (Table 1) and used as input of secretome analysis.
Table 1. Total of Augustus predictions obtained for 5 fungi models organisms.
The third column shows the number of sequences >70 aa used in secretome analysis.

Secretome candidates


Searching for transporter signals

We included 2888 predicted sequences in TargetP 1.1 software that uses neural networks (NN) and hidden Markov models (HMM) trained on eukaryotes (non-plant organisms).

Prediction resulted in a total of 203 sequences with high Mithocondrial scores in relation to other HMM prediction for Peptide Signal and other transported signals. Table 2 shows 31 sequences with score cutoff of > 0.80. For detailed information of TargetP analysis click on OutputTargetP.

Table 2. Predicted sequences with HMM Mithocondrial transported probability > 0.8 after TargetP analysis.


TargetP provides a prediction for signal peptide too. The prediction results on 154 sequences with a score cutoff for SignalP higher than Mithocondrial Peptide and others peptides (OutputTargetSP) This sequences were selected to run SignalP Server and obtain deep information about the prediction.

Searching for putative extracellular proteins

We selected 154 predicted sequences by TargetP to run SignalP 3.0 Server . Table 3 shows our candidate secretome sequences filtered using a HMM probability > 0.9. For total dates of SignalP prediction click on OutputSignalP

Table 3. Candidate secretome sequences of coffe rust genome identified using SignalP model.