Improved prediction of fungal effector proteins from secretomes with EffectorP 2.0

Abstract

4Plant-pathogenic fungi secrete effector proteins to facilitate infection. We describe extensive improvements to 1 5EffectorP, the first machine learning classifier for fungal effector prediction. EffectorP 2.0 is now trained on a larger 1 6 set of effectors and utilizes a different approach based on an ensemble of classifiers trained on different subsets of 1 7 negative data, offering different views on classification. EffectorP 2.0 achieves accuracy of 89%, compared to 82% for 1 8 EffectorP 1.0 and 59.8% for a small size classifier. Important features for effector prediction appear to be protein size, 1 9protein net charge as well as the amino acids serine and cysteine. EffectorP 2.0 decreases the number of predicted 2 0 effectors in secretomes of fungal plant symbionts and saprophytes by 40% when compared to EffectorP 1.0. However, 2 1 EffectorP 1.0 retains value and combining EffectorP 1.0 and 2.0 results in a stringent classifier with low false positive 2 2 rate of 9%. EffectorP 2.0 predicts significant enrichments of effectors in 12 out of 13 sets of infection-induced 2 3proteins from diverse fungal pathogens, whereas a small cysteine-rich classifier detects enrichment only in 7 out of 13. 2 4 EffectorP 2.0 will fast-track prioritization of high-confidence effector candidates for functional validation and aid in 2 5 improving our understanding of effector biology. EffectorP 2.0 is available at http://effectorp.csiro.au. 2 6

Comparative analyses of saprotrophy in Salisapilia sapeloensis and diverse plant pathogenic oomycetes reveal lifestyle-specific gene expression

“…We therefore used a heuristic and HMM search to identify initial candidates for RxLRs and CRN. For SSPs we used EffectorP2.0 [52]. Because these are secreted, cytoplasmic effectors we further determined candidate genes without a transmembrane domain (TM domain) and with a SP.…”

Time‐resolved dissection of the molecular crosstalk driving Fusarium head blight in wheat provides new insights into host susceptibility determinism

“…Subcellular localization was predicted using WoLF PSORT v0.1 (Horton et al, ), specifying “plant organism” for secreted predicted proteins and “fungi organism” for those not predicted as part of the secretome. Putative effectors were traced using the EffectorP v2.0 tool with a score >0.55 (Sperschneider, Dodds, Gardiner, Singh, & Taylor, ). GO annotation was performed using HMMER 3.1b2 (February 2015) by comparison of protein sequences with HMM PFAM‐A database (01/2017; P value <0.01; dom‐evalue <0.01).…”

Novel regulators of nitric oxide signaling triggered by host perception in a plant pathogen

“…Secretome of Fg was inferred from a previous report 73 . Putative effector proteins were predicted using EffectorP2 with default settings 74 . To identify secreted peptidases, sequences of the predicted secreted proteins were subjected to a local BLASTP database based on MEROPS 75 .…”