Next Article in Journal
A Luciferase Immunosorbent Assay Based on Attachment Glycoprotein for the Rapid and Easy Detection of Nipah Virus IgG Antibodies
Previous Article in Journal
Intestinal Emphysema and Gut Bacterial Microbiota Composition
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Microorganisms
Volume 12
Issue 5
10.3390/microorganisms12050982
microorganisms-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles thumb_up
...
Endorse textsms
...
Comment
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Machine Learning-Based Identification of Mating Type and Metalaxyl Response in Phytophthora infestans Using SSR Markers

by
Collins A. Agho
1,*,
Jadwiga Śliwka
2,
Helina Nassar
1,
Ülo Niinemets
1,3,* and
Eve Runno-Paurson
1
1
Institute of Agricultural and Environmental Sciences, Estonian University of Life Sciences, Kreutzwaldi 1, 51006 Tartu, Estonia
2
Plant Breeding and Acclimatization Institute—National Research Institute in Radzików, Department of Potato Genetics and Parental Lines, Platanowa Str. 19, 05-831 Młochów, Poland
3
Estonian Academy of Sciences, Kohtu 6, 10130 Tallinn, Estonia
*
Authors to whom correspondence should be addressed.
Microorganisms 2024, 12(5), 982; https://doi.org/10.3390/microorganisms12050982
Submission received: 18 March 2024 / Revised: 6 May 2024 / Accepted: 9 May 2024 / Published: 14 May 2024
(This article belongs to the Section Systems Microbiology)
Browse Figures
Versions Notes

Abstract

:
Phytophthora infestans is the causal agent of late blight in potato. The occurrence of P. infestans with both A1 and A2 mating types in the field may result in sexual reproduction and the generation of recombinant strains. Such strains with new combinations of traits can be highly aggressive, resistant to fungicides, and can make the disease difficult to control in the field. Metalaxyl-resistant isolates are now more prevalent in potato fields. Understanding the genetic structure and rapid identification of mating types and metalaxyl response of P. infestans in the field is a prerequisite for effective late blight disease monitoring and management. Molecular and phenotypic assays involving molecular and phenotypic markers such as mating types and metalaxyl response are typically conducted separately in the studies of the genotypic and phenotypic diversity of P. infestans. As a result, there is a pressing need to reduce the experimental workload and more efficiently assess the aggressiveness of different strains. We think that employing genetic markers to not only estimate genotypic diversity but also to identify the mating type and fungicide response using machine learning techniques can guide and speed up the decision-making process in late blight disease management, especially when the mating type and metalaxyl resistance data are not available. This technique can also be applied to determine these phenotypic traits for dead isolates. In this study, over 600 P. infestans isolates from different populations—Estonia, Pskov region, and Poland—were classified for mating types and metalaxyl response using machine learning techniques based on simple sequence repeat (SSR) markers. For both traits, random forest and the support vector machine demonstrated good accuracy of over 70%, compared to the decision tree and artificial neural network models whose accuracy was lower. There were also associations (p < 0.05) between the traits and some of the alleles detected, but machine learning prediction techniques based on multilocus SSR genotypes offered better prediction accuracy.

1. Introduction

Potatoes (Solanum tuberosum, Solanaceae) are one of the most significant vegetables worldwide [1]. Late blight Phytophthora infestans (P. infestans) is a pathogen that affects field-grown potatoes and tomatoes all over the world and has historically been common in many countries [2,3,4,5,6,7,8]. Throughout the growing season, P. infestans can cause water-soaked lesions on the plant leaves, stems, and fruits, with an overall effect of reducing yield quality and quantity [9,10]. It is the most devastating disease to this crop, posing a threat to global food security [10,11,12].
A better understanding of the population dynamics in P. infestans has been made possible by diversity studies that take into account both phenotypic and genotypic markers [6,13,14,15,16]. Mating type, virulence spectrum, and metalaxyl/mefenoxam resistance are frequent phenotypic markers [17,18]. Molecular markers such as simple sequence repeat (SSR) markers, amplified fragment length polymorphism (AFLP), and random fragment length polymorphism (RFLP) are commonly employed in P. infestans studies due to their high repeatability [6,17,19]. P. infestans has two mating types, A1 and A2, and it can reproduce both asexually and sexually. Sexual reproduction can occur when both mating types are present [20]. With the exception of Mexico, which was thought to be the center of origin where the A1 and A2 mating types co-exist, only the A1 mating type isolates were known to occur until the early 1980s [21]. However, as the A2 mating type spread throughout Europe and other parts of the world, sexual reproduction became more common, increasing the genetic diversity and adaptability of the pathogen, resulting in the formation of oospores that can survive for a long time in the soil, even during the winter, and can infect other plants the following season [21,22,23,24,25].
Since the 1970s, phenylamide fungicides (initially, the racemic mixture of metalaxyl was used prior to the introduction of its biologically active enantiomer R-metalaxyl (also known as mefenoxam)) have made substantial contributions to the efficient management of P. infestans [6,12,18,26,27]. However, fungicide resistance to late blight is a problem that is getting worse due to the introduction of novel strains that are tolerant or resistant to metalaxyl. As early as the 1980s, reports of P. infestans isolates in Europe that are resistant to metalaxyl were published [28,29]. The resistance can result from inactivation, altered transport, altered metabolism of the fungicide or modification of the fungicide target as a result of pathogen mutations [30,31,32]. Frequent use of fungicides can create fungicide selection pressure, which can lead to a shift from sensitivity to resistance over time [33,34]. Though the molecular mechanism of fungicide resistance is still not fully understood [12,26], some DNA markers linked to this trait have been suggested [26,35,36]. Ref. [37] reported a complicated inheritance pattern involving both major and minor loci for metalaxyl insensitivity in P. infestans, indicating that metalaxyl resistance is controlled by multiple loci.
The conventional method for determining P. infestans response to metalaxyl relies on a standard microbiological assay, in which control and fungicide-amended media are inoculated with the P. infestans and the growth of the colonies is measured and compared after 7–14 days [15,38]. To determine the mating types, matching an unknown isolate with tester isolates on an agar plate and watching for oospore formation is used [39]. Despite being straightforward, this method necessitates the isolation and use of live P. infestans strains [40]. This method can be time-consuming, requiring 10 to 14 days to ascertain oospore formation, and it prevents the determination of mating-type information on dead biological materials [40].
Molecular and phenotypic assays involving molecular and phenotypic markers such as mating types and metalaxyl response are typically conducted separately in the studies of the genotypic and phenotypic diversity of P. infestans. It is critical to have a faster methodology that includes swift phenotyping for fungicide response and mating type, while also saving cost when assessing P. infestans populations for diversity studies in the event of a potential outbreak or study of the pathogen dynamics, considering its devastating nature [41]. We argue that molecular markers can be utilized to assess not only genotypic diversity but the same marker profile can also be used to determine the mating type and fungicide response using machine learning techniques. This will allow for a quicker analysis of genotypic and phenotypic changes in the population of P. infestans, and the technology could lead to optimum management strategies [12], especially when mating type and metalaxyl resistance data are not available, such as in the case of dead isolates. Molecular markers, including SSR markers, have been used for pathogen differentiation, genomic prediction of resistance to pathogens, and characterization of various organisms [42,43,44,45,46]. In previous studies, the use of SSR, inter-simple sequence repeat (ISSR), and sequence-related amplified polymorphism (SRAP) markers gave a good classification for mefenoxam susceptibility and pathotype classification in Pseudoperonospora cubensis (P. cubensis) [43]. Machine learning techniques have been helpful, with reasonable accuracy, in reducing the workload for modeling and prediction, and molecular and genetic characterization in microbiological and agricultural applications [43,46,47]. For example, notable accuracy was achieved with a particle swarm optimization (PSO) trained artificial neural network (ANN) in the identification of fungicide resistance levels in P. cubensis [43]. For cultivar characterization in potato, ANN was also relatively accurate [47]. In another study, the support vector machine (SVM) and Naive Bayes machine learning models were employed to accurately identify cultivars of olive (Olea europaea) trees based on Random Amplified Polymorphic DNA (RAPD) and ISSR genetic markers [48].
The application of machine learning techniques to empirical datasets generated from population genetics experiments seems to have received less attention than it may have deserved, given its practical effectiveness [46]. Sexual reproduction is crucial in the biology and epidemiology of P. infestans. P. infestans populations with A1 and A2 mating types have the capacity to undergo sexual reproduction. This can enhance their longevity and evolution of strains with higher virulence and resistance to common fungicides such as metalaxyl [49]. Therefore, a quick genetic test to assess the prevalence of A1 and A2 mating types, and fungicide-resistant isolates in field populations of P. infestans can reduce the workload associated with microbiological assays, and save time and cost, especially in disease monitoring and management. DNA markers such as AFLP markers have been used to identify mating types [49], and metalaxyl resistance [35]. P. infestans fungicide resistance has also been determined using Fourier Transform Infra-Red spectroscopy (FTIR) [12]. However, there are no machine learning approaches for mating type and metalaxyl response identification using SSR markers. Our objective was to assess the general efficiency of SSR markers with machine learning approaches and evaluate the efficiency of different machine learning methods in classifying mating type and fungicide response in diverse populations of P. infestans.

2. Materials and Methods

SSR markers were used to characterize P. infestans isolates from Estonia, Pskov region (Russia) (published previously by [6,27], and Poland (published previously by [50,51]. These isolates were genotyped based on the following 12 SSR markers: D13, G11, Pi04, Pi4B, Pi63, Pi70, SSR2, Pi02, SSR4, SSR6, SSR8, and SSR11. The isolates were also phenotyped for mating type (A1 and A2 mating types) and metalaxyl response (resistance, intermediate, and sensitive). For mating type, a total of 850 isolates consisting of 141 isolates from the Pskov region, 111 isolates from Estonia, and 598 isolates from Poland were used. In total, there were 566 A1 and 284 A2 mating types. For metalaxyl response, a total of 823 isolates consisting of 120 isolates from the Pskov region, 111 isolates from Estonia, and 592 isolates from Poland were used. In total, there were 122 isolates of intermediate response, 126 that were resistant, and 575 that were sensitive to metalaxyl.

3. Data Preparation

The data were clone-corrected for mating type and metalaxyl resistance using the package poppr [52,53] in R (ver. 4.2.0, R Core Team, Vienna, Austria, 2022 [54]). The R package poppr was also used to plot a genotype accumulation curve and assess the minimum number of loci that can sufficiently differentiate between different genotypes in the P. infestans population [53]. The curve shows a greatly decreased variance over 12 SSR markers and correctly identifies 100% of the multilocus genotypes, indicating its sufficiency for complete genotyping (Figure 1a,b). After correction to remove duplicate genotypes, 629 isolates for mating type and 617 isolates for metalaxyl response remained.
In total, there were 99 isolates with intermediate response, 105 isolates were resistant, and 413 isolates were sensitive to metalaxyl, respectively. For mating types, 403 isolates were of the A1 mating type and 226 isolates were of the A2 mating type. For predictions, two comparisons were made where one involves using all the alleles for classification and the other uses only common alleles among the populations (Estonia, Pskov region, and Poland). There were 105 alleles in total for both traits (Table S1). Alleles that were specific to each of the populations were removed from the dataset, leaving only the common alleles among the populations. In total, there were 32 common alleles among Estonia, Pskov region, and Poland P. infestans isolates for mating type and 31 common alleles for metalaxyl response. Alleles were recoded as 0 for the absence of the allele, 1 for the presence of one copy of the allele, or 2 for the presence of two copies of the allele per locus.

4. Machine Learning Methods

The artificial neural network (ANN), support vector machine (SVM), random forest, and decision tree classifier were used as model approaches in this study. Multilayer perceptron (MLP) is a straightforward, deep, feed-forward artificial neural network with three layers (the input, hidden, and output layers), and the neurons of a layer are fully connected to the neurons of the next layers [55]. Each neuron determines its output signal by applying an activation function to its net input, which is the sum of weighted input signals [56]. The artificial neural network (ANN) is an extension of the MLP with more than one hidden layer [57]. ANNs, without having to explicitly explain the complex relationships in the feature space, as required by classical statistical techniques, can tackle non-linear complex classification challenges [58]. This study uses an ANN consisting of the input layer that holds data entering the neural network, three hidden layers that perform nonlinear transformations of the inputs entered into the network, and finally, an output layer that produces the output variables [59,60]. The support vector machine approach is based on statistical learning theory and provides a powerful tool to address the problem of classification [56]. Within two classes of data, SVM determines the hyperplane with the increased margin [57]. Its improved generalization performance is based on the Structural Risk Minimization (SRM) principle [56]. On the other hand, random forest is simple, yet effective in classification and regression, and uses multiple decision trees during the classification process to obtain more accurate results [61]. Random forest creates a model to explain the classification by the predictor variables (traits) using a training subset of the original dataset (in this case, isolates) [61]. A decision tree is a classification method that uses a set of tests that are established at each branch (or node) in the tree to recursively divide a dataset into smaller groups [62]. Since it is strictly nonparametric, no assumptions about the distributions of the input data are necessary [62], and it is a dominant and acceptable method for classification and prediction [57]. For both traits (mating type and metalaxyl response), 70% of the dataset was used as the training set, while the remaining 30% was used as the testing set. The random forest, decision tree classifier, support vector machine, and ANN were implemented using Python modules within the Jupyter Notebook interface. For the ANN, three hidden layers were used with a depth of four (three hidden layers, and one output layer, excluding the input layer) [60]. The activation function sigmoid was used as the output layer for the mating type (binary outcome), while for the metalaxyl response (more than two categories), the activation function softmax was used as the output layer [60]. Accuracy of performance was accessed using classification accuracy (CA). It is a statistic that is frequently used to gauge how well machine learning algorithms are performing, defined by the number of correct predictions over the number of total predictions [43,46]. Precision, recall, and the F1 score were also computed. Recall is the proportion of correctly classified positive predictions among all true positive predictions; precision is the proportion of positively classified predictions that are true positives; and the F1 score is the harmonic mean of these two measures, and ranges from 0 to 1 [63]. A higher F1 score (closer to 1) indicates a better model. The performance metrics were calculated according to [64]. The associations of the alleles with the traits were also tested based on logistic regression and stepwise discriminant analysis at p < 0.05. Due to the considerably large training data for the mating type, metalaxyl response, and hyperparameter tuning, the overfitting problem for all the models was not a severe issue [65].

5. Results

Among the P. infestans isolates characterized with SSR markers, 105 alleles were detected at 12 loci, all of which were polymorphic (Figure 2). They displayed allelic diversity among the isolates, with 3 (SSR2 and Pi70) to 21 alleles (G11) per locus, with an average of 8.75 ± 1.64 alleles.
The classification accuracy based on random forest, decision tree classifier, support vector machine, and artificial neural network for the mating type was 76.7%, 68.3%, 70.4%, and 37.6%, respectively, using all alleles (Figure 3b). For the metalaxyl response, classification accuracy based on random forest, decision tree classifier, support vector machine, and ANN for the metalaxyl response was 75.8%, 68.3, 73.7%, and 58.6%, respectively, using all alleles (Figure 4b). Using only common alleles, the classification accuracy based on random forest, decision tree classifier, support vector machine, and artificial neural network for the mating type was 75.7%, 67.2%, 67.7%, and 64.6%, respectively (Figure S1). For the metalaxyl response, classification accuracy using only common alleles based on random forest, decision tree classifier, support vector machine, and ANN for the metalaxyl response was 74.2%, 72.0%, 73.7%, and 63.4%, respectively (Figure S1). For all the machine learning models, random forest consistently gave a higher classification accuracy (typically above 75%) for both traits, followed by SVM (above 70%) using all alleles (Table 1). Likewise, using only common alleles, random forest consistently gave a higher classification accuracy (above 70%) for both traits (Table 1). This higher prediction for random forest is also reflected in the F1 score, which was consistently higher, compared to other models for both traits (Table 1). Compared to the other machine learning models, ANN ranked lower for classification accuracy for both traits when either all marker alleles or only common marker alleles were used. A total of 17 out of the 105 alleles were significantly (p < 0.05) associated with metalaxyl response (D13.136, D13.162, G11.160, G11.198, G11.164, G11.152, G11.204, Pi4B.205, Pi63.273, Pi63.151, SSR2.173, Pi02.258, Pi02.266, SSR4.292, SSR4.297, SSR6.232, SSR8.266). Only four alleles were significantly (p < 0.05) associated with mating types (G11.156, G11.202, G11.198, and Pi04.160).

6. Discussion

Machine learning has found applications in agriculture, including marker–trait association and genomic prediction [66,67,68]. Molecular marker technology has advanced our understanding in plant pathology research [69]. Among different classes of molecular markers, SSR markers have been widely used in numerous applications in plant pathology, such as genotyping pathogen populations, genetic diversity studies, monitoring changes in pathogen populations, phylogeny, parentage analyses, and reproductive biology of various plant pathogens [6,70,71,72,73]. This can be attributed to their multi-allelic nature, co-dominant inheritance (an important feature when analyzing the populations of diploid organisms such as P. infestans), relative abundance, repeatability, good genome coverage [74,75], and genotyping, which can be quickly completed by multiplexing all the pairs of SSR primers in a single reaction [76]. Indeed, in this study, all the loci were polymorphic, with most alleles found in the loci D13, SSR4, and G11, similar to those obtained by [77] in P. infestans genetic diversity studies using SSR markers.
In this study, different machine learning methods were applied in classifying mating type and metalaxyl fungicide response in P. infestans using SSR markers. The high classification accuracy across the models, except for the ANN that was considerably lower, indicates that these phenotypic traits likely have a genetic basis, and can be predicted using markers. Indeed, previous studies have linked DNA markers to metalaxyl response [26,35,36] and mating types [26,49], and several significant single nucleotide polymorphism markers (SNPs) have been associated with mating type and fungicide response in Phytophthora spp. [78]. The accuracy of the metalaxyl-sensitive response prediction was very high, ranging from 82.8% to 97.5%, compared to the prediction for other classes. The prediction of the A1 mating type was also more accurate compared to the prediction for the A2 mating type for all models except the ANN. This may be due to a larger training dataset for these classes. Ref. [36] observed that the allele 195 in locus Pi70 and the allele 166 in locus Pi04, which were common for all P. infestans populations in our study, were linked to metalaxyl-M (Mefenoxam) response. In our study, these alleles were not significantly associated with metalaxyl response. A total of 17 out of the 105 alleles were significantly (p < 0.05) associated with metalaxyl response. Comparably, several loci have been linked to insensitivity to metalaxyl in P. infestans [37,79]. However, only a few alleles were significantly associated with mating types, mainly in the G11 locus, indicating that mating types may be associated with a few major loci.
One would expect the ANN to perform better than other models due to the model architecture that closely mimics the human brain and its suitability for complex classification [80]. Yet, random forest was the best classifier for mating type and metalaxyl response. ANN models are more difficult to parametrize, and the choice of model parameters such as activation function, number of neurons, and number of hidden layers can greatly influence its performance. ANN has been shown to be accurate enough for the identification of pathogens and fungicide resistance and is thus suitable for use in practical situations [43]. Unlike the decision tree classifier, random forest as an ensemble method has greater accuracy, which can be attributed to its robustness in the selection of training samples and noise in the training dataset [81].
This study attempted to compare machine learning models for the classification of mating type and metalaxyl response in P. infestans using SSR markers. Although the classification accuracy can be deemed adequate in the machine learning context, particularly for the random forest model, it is debatable whether the accuracy is sufficient to guide decision-making. Moreover, in the plants within some machine learning studies, such as for the prediction of miRNAs associated with abiotic stresses [82], an accuracy of less than 70% was reported for some models. In humans, an accuracy of 65% to 70% was reported for diagnostic prediction of autism [83]. Even for genomic selection where markers are generally regarded as linked to phenotypes, an accuracy of less than 70% is common [84]. Thus, anything greater than 70% accuracy can be considered promising and has been reported, even with single nucleotide polymorphism markers [85,86,87,88]. However, our study shows a statistically significant association between the SSR marker alleles and metalaxyl response/mating type phenotypes, which opens up the possibility for the use of a slow-evolving SNP-based markers for the identification of both traits. On the other hand, incorporating more SSR markers and more datasets to balance the ratio of A1 and A2 mating types, as well as ratios of resistance and the intermediate and sensitive metalaxyl response in the training set, could also improve the quality of the dataset and thus the accuracy of models. Further hyperparameter optimization of the models can improve model performance and reduce any obvious overfitting. It can also be useful to examine the effect of decreasing or increasing the complexity of the models by feature engineering. Also, as high-throughput technology improves and as more markers are developed, a much higher predictive accuracy will likely be achieved. This kind of prediction could be a valuable addition to P. infestans population diversity studies based on SSR markers, aiding in the assessment of the metalaxyl fungicide effectiveness over time and space, and the possibility of sexual reproduction in the field (which can change the epidemiology of potato late blight and, in turn, changes the disease management strategies), especially when mating type and metalaxyl resistance data are not available. To ascertain the mating type and metalaxyl response, all that has to be done is input the unknown isolate’s SSR profile into a training model that has been developed using the SSR marker data training set. The program will then take up to a few seconds or minutes to execute its function. This technique may be used on both live and dead or archived biological samples, which makes it a useful addition to the toolkit for identifying mating type and metalaxyl response in P. infestans populations.

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/microorganisms12050982/s1. Figure S1: Metalaxyl response and mating type accuracy for the different machine learning models using common SSR marker alleles; Table S1: Changes in allele frequency of SSR marker alleles of P. infestans isolates, for common and all alleles across populations for mating type and metalaxyl response.

Author Contributions

C.A.A. conceptualized the study concept and its design, analyzed, and interpreted the data, wrote the original draft of the manuscript, and critically revised and edited the manuscript for important intellectual content. E.R.-P. critically examined the study concept and design, and reviewed and edited the manuscript. J.Ś. supplied the datasets from the Polish P. infestans population, reviewed and edited the manuscript. H.N. edited the manuscript. Ü.N. coordinated and supervised the research, critically examined the study concept and design, guided the discussion of the outcomes, and revised and edited the manuscript for important intellectual content. All authors have read and agreed to the published version of the manuscript.

Funding

The study was funded by the Estonian Ministry of Education and Research (Center of Excellence AgroCropFuture “Agroecology and new crops in future climates”, TK200), the Estonian University of Life Sciences base funding project P190259PKTT, and Estonian Research Council (PRG2207). The research equipment used in the project was partly funded by the Estonian Research Council project „Plant Biology Infrastructure—TAIM“ (TT5).

Data Availability Statement

The original contributions presented in the study are included in the article and Supplementary Material, further inquiries can be directed to the corresponding author.

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. Birch, P.R.; Bryan, G.; Fenton, B.; Gilroy, E.M.; Hein, I.; Jones, J.T.; Prashar, A.; Taylor, M.A.; Torrance, L.; Toth, I.K. Crops that feed the world 8: Potato: Are the trends of increased global production sustainable? Food Secur. 2012, 4, 477–508. [Google Scholar] [CrossRef]
  2. Cui, J.; Luan, Y.; Jiang, N.; Bao, H.; Meng, J. Comparative transcriptome analysis between resistant and susceptible tomato allows the identification of lnc RNA 16397 conferring resistance to Phytophthora infestans by co-expressing glutaredoxin. Plant J. 2017, 89, 577–589. [Google Scholar] [CrossRef] [PubMed]
  3. El-Ganainy, S.M.; Iqbal, Z.; Awad, H.M.; Sattar, M.N.; Tohamy, A.M.; Abbas, A.O.; Squires, J.; Cooke, D.E. Genotypic and phenotypic structure of the population of Phytophthora infestans in Egypt revealed the presence of European genotypes. J. Fungi 2022, 8, 468. [Google Scholar] [CrossRef] [PubMed]
  4. Kim, M.J.; Mutschler, M.A. Transfer to processing tomato and characterization of late blight resistance derived from Solanum pimpinellifolium L. L3708. J. Am. Soc. Hortic. Sci. 2005, 130, 877–884. [Google Scholar] [CrossRef]
  5. Miranda, B.E.C.; Suassuna, N.D.; Reis, A. Mating type, mefenoxam sensitivity, and pathotype diversity in Phytophthora infestans isolates from tomato in Brazil. Pesqui. Agropecu. Bras. 2010, 45, 671–679. [Google Scholar] [CrossRef]
  6. Runno-Paurson, E.; Agho, C.A.; Zoteyeva, N.; Koppel, M.; Hansen, M.; Hallikma, T.; Cooke, D.E.; Nassar, H.; Niinemets, Ü. Highly diverse Phytophthora infestans populations infecting potato crops in Pskov region, North-West Russia. J. Fungi 2022, 8, 472. [Google Scholar] [CrossRef]
  7. Runno-Paurson, E.; Nassar, H.; Tähtjärv, T.; Eremeev, V.; Hansen, M.; Niinemets, Ü. High temporal variability in late blight pathogen diversity.; virulence.; and fungicide resistance in potato breeding fields: Results from a long-term monitoring study. Plants 2022, 11, 2426. [Google Scholar] [CrossRef]
  8. Śliwka, J.; Sobkowiak, S.; Lebecka, R.; Avendaño-Córcoles, J.; Zimnoch-Guzowska, E. Mating type, virulence, aggressiveness and metalaxyl resistance of isolates of Phytophthora infestans in Poland. Potato Res. 2006, 49, 155–166. [Google Scholar] [CrossRef]
  9. Fry, W.E.; Goodwin, S.B.; Matuszak, J.M.; Spielman, L.J.; Milgroom, M.G.; Drenth, A. Population genetics and intercontinental migrations of Phytophthora infestans. Annu. Rev. Phytopathol. 1992, 30, 107–130. [Google Scholar] [CrossRef]
  10. Park, Y.; Hwang, J.; Kim, K.; Kang, J.; Kim, B.; Xu, S.; Ahn, Y. Development of the gene-based SCARs for the Ph-3 locus, which confers late blight resistance in tomato. Sci. Hortic. 2013, 164, 9–16. [Google Scholar] [CrossRef]
  11. Adolf, B.; Andrade-Piedra, J.; Bittara Molina, F.; Przetakiewicz, J.; Hausladen, H.; Kromann, P.; Lees, A.; Lindqvist-Kreuze, H.; Perez, W.; Secor, G.A.; et al. Fungal, oomycete, and plasmodiophorid diseases of potato. In The Potato Crop: Its Agricultural, Nutritional and Social Contribution to Humankind; Springer: Cham, Switzerland, 2020; pp. 307–350. [Google Scholar]
  12. Pomerantz, A.; Cohen, Y.; Shufan, E.; Ben-Naim, Y.; Mordechai, S.; Salman, A.; Huleihel, M. Characterization of Phytophthora infestans resistance to mefenoxam using FTIR spectroscopy. J. Photochem. Photobiol. B Biol. 2014, 141, 308–314. [Google Scholar] [CrossRef] [PubMed]
  13. Day, J.P.; Wattier, R.A.M.; Shaw, D.S.; Shattock, R.C. Phenotypic and genotypic diversity in Phytophthora infestans on potato in Great Britain, 1995–1998. Plant Pathol. 2004, 53, 303–315. [Google Scholar] [CrossRef]
  14. Hansen, Z.R.; Everts, K.L.; Fry, W.E.; Gevens, A.J.; Grünwald, N.J.; Gugino, B.K.; Johnson, D.A.; Johnson, S.B.; Judelson, H.S.; Knaus, B.J.; et al. Genetic variation within clonal lineages of Phytophthora infestans revealed through genotyping-by-sequencing, and implications for late blight epidemiology. PLoS ONE 2016, 11, e0165690. [Google Scholar] [CrossRef] [PubMed]
  15. Hermansen, A.; Hannukkala, A.; Naerstad, R.H.; Brurberg, M.B. Variation in populations of Phytophthora infestans in Finland and Norway: Mating type.; metalaxyl resistance and virulence phenotype. Plant Pathol. 2000, 49, 11–22. [Google Scholar] [CrossRef]
  16. Knapova, G.; Gisi, U. Phenotypic and genotypic structure of Phytophthora infestans populations on potato and tomato in France and Switzerland. Plant Pathol. 2002, 51, 641–653. [Google Scholar] [CrossRef]
  17. Blandón-Díaz, J.U.; Widmark, A.K.; Hannukkala, A.; Andersson, B.; Högberg, N.; Yuen, J.E. Phenotypic variation within a clonal lineage of Phytophthora infestans infecting both tomato and potato in Nicaragua. Phytopathology 2012, 102, 323–330. [Google Scholar] [CrossRef] [PubMed]
  18. Göre, M.E.; Altın, N.; Myers, K.; Cooke, D.E.; Fry, W.E.; Özer, G. Population structure of Phytophthora infestans in Turkey reveals expansion and spread of dominant clonal lineages and virulence. Plant Pathol. 2021, 70, 898–911. [Google Scholar] [CrossRef]
  19. Cooke, D.E.L.; Lees, A.K. Markers, old and new for examining Phytophthora infestans diversity. Plant Pathol. 2004, 53, 692–704. [Google Scholar] [CrossRef]
  20. Goodwin, S.B.; Sujkowski, L.S.; Dyer, A.T.; Fry, B.A.; Fry, W.E. Direct detection of gene flow and probable sexual reproduction of Phytophthora infestans in northern North America. Phytopathology 1995, 85, 473–479. [Google Scholar] [CrossRef]
  21. Fry, W.E.; Goodwin, S.B.; Dyer, A.T.; Matuszak, J.M.; Drenth, A.; Tooley, P.W.; Sujkowski, L.S.; Koh, Y.J.; Cohen, B.A.; Spielman, L.J.; et al. Historical and recent migrations of Phytophthora infestans: Chronology, pathways and implications. Plant Dis. 1993, 77, 653–661. [Google Scholar] [CrossRef]
  22. Cooke, L.R.; Schepers, H.T.; Hermansen, A.; Bain, R.A.; Bradshaw, N.J.; Ritchie, F.; Shaw, D.S.; Evenhuis, A.; Kessel, G.J.; Wander, J.G.; et al. Epidemiology and integrated control of potato late blight in Europe. Potato Res. 2011, 54, 183–222. [Google Scholar] [CrossRef]
  23. Drenth, A.; Janssen, E.M.; Govers, F. Formation and survival of oospores of Phytophthora infestans under natural conditions. Plant Pathol. 1995, 44, 86–94. [Google Scholar] [CrossRef]
  24. McDonald, B.A.; Linde, C. Pathogen population genetics, evolutionary potential, and durable resistance. Annu. Rev. Phytopathol. 2002, 40, 349–379. [Google Scholar] [CrossRef] [PubMed]
  25. Santana, F.M.; Gomes, C.B.; Rombaldi, C.; Bianchi, V.J.; Reis, A. Characterization of Phytophthora infestans populations of southern Brazil in 2004 and 2005. Phytoparasitica 2013, 41, 557–568. [Google Scholar] [CrossRef]
  26. Ayala-Usma, D.A.; Danies, G.; Myers, K.; Bond, M.O.; Romero-Navarro, J.A.; Judelson, H.S.; Restrepo, S.; Fry, W.E. Genome-wide association study identifies single nucleotide polymorphism markers associated with mycelial growth (at 15, 20, and 25 C), mefenoxam resistance, and mating type in Phytophthora infestans. Phytopathology 2020, 110, 822–833. [Google Scholar] [CrossRef] [PubMed]
  27. Runno-Paurson, E.; Agho, C.A.; Nassar, H.; Hansen, M.; Leitaru, K.; Hallikma, T.; Cooke, D.E.L.; Niinemets, Ü. The variability of Phytophthora infestans isolates collected from Estonian islands in the Baltic Sea. Plant Dis. 2024. [Google Scholar] [CrossRef] [PubMed]
  28. Davidse, L.C.; Danial, D.L.; Van Westen, C.J. Resistance to metalaxyl in Phytophthora infestans in the Netherlands. Neth. J. Plant Pathol. 1983, 89, 1–20. [Google Scholar] [CrossRef]
  29. Dowley, L.J.; O’sullivan, E. Metalaxyl-resistant strains of Phytophthora infestans (Mont.) de Bary in Ireland. Potato Res. 1981, 24, 417–421. [Google Scholar] [CrossRef]
  30. Hahn, M. The rising threat of fungicide resistance in plant pathogenic fungi: Botrytis as a case study. J. Chem. Biol. 2014, 7, 133–141. [Google Scholar] [CrossRef]
  31. Lee, T.Y.; Mizubuti, E.; Fry, W.E. Genetics of metalaxyl resistance in Phytophthora infestans. Fungal Genet. Biol. 1999, 26, 118–130. [Google Scholar] [CrossRef]
  32. Steffens, J.J.; Pell, E.J.; Tien, M. Mechanisms of fungicide resistance in phytopathogenic fungi. Curr. Opin. Biotechnol. 1996, 7, 348–355. [Google Scholar] [CrossRef] [PubMed]
  33. Adaskaveg, J.E.; Michailides, T.; Eskalen, A. Fungicides, Bactericides, Biocontrols, and Natural Products for Deciduous Tree Fruit and Nut, Citrus, Strawberry, and Vine Crops in California; University of California: Davis, CA, USA, 2022. [Google Scholar]
  34. Pliakhnevich, M.; Ivaniuk, V. Aggressiveness and metalaxyl sensitivity of Phytophthora infestans strains in Belarus. Zemdirbyste 2008, 95, 379–387. [Google Scholar]
  35. Eom, S.H.; Kim, K.J.; Jung, H.S.; Lee, S.P.; Lee, Y.S. Identification of DNA Markers Linked to Metalaxyl Insensitivity Loci in Phytophthora infestans. Mycobiology 2003, 31, 229–234. [Google Scholar] [CrossRef]
  36. Montes, M.S.; Nielsen, B.J.; Schmidt, S.G.; Bødker, L.; Kjøller, R.; Rosendahl, S. Population genetics of Phytophthora infestans in Denmark reveals dominantly clonal populations and specific alleles linked to metalaxyl-M resistance. Plant Pathol. 2016, 65, 744–753. [Google Scholar] [CrossRef]
  37. Fabritius, A.L.; Shattock, R.C.; Judelson, H.S. Genetic analysis of metalaxyl insensitivity loci in Phytophthora infestans using linked DNA markers. Phytopathology 1997, 87, 1034–1040. [Google Scholar] [CrossRef] [PubMed]
  38. Saville, A.; Graham, K.; Grünwald, N.J.; Myers, K.; Fry, W.E.; Ristaino, J.B. Fungicide sensitivity of US genotypes of Phytophthora infestans to six oomycete-targeted compounds. Plant Dis. 2015, 99, 659–666. [Google Scholar] [CrossRef]
  39. Runno-Paurson, E.; Fry, W.E.; Myers, K.L.; Koppel, M.; Mänd, M. Characterisation of Phytophthora infestans isolates collected from potato in Estonia during 2002–2003. Eur. J. Plant Pathol. 2009, 124, 565–575. [Google Scholar] [CrossRef]
  40. Mabon, R.; Guibert, M.; Corbière, R.; Andrivon, D. An improved PCR method for rapid and accurate identification of mating types in the late blight pathogen Phytophthora infestans. Plant Health Prog. 2021, 22, 362–367. [Google Scholar] [CrossRef]
  41. Raffaele, S.; Win, J.; Cano, L.M.; Kamoun, S. Analyses of genome architecture and gene expression reveal novel candidate virulence factors in the secretome of Phytophthora infestans. BMC Genom. 2010, 11, 637. [Google Scholar] [CrossRef]
  42. Fraslin, C.; Koskinen, H.; Nousianen, A.; Houston, R.D.; Kause, A. Genome-wide association and genomic prediction of resistance to Flavobacterium columnare in a farmed rainbow trout population. Aquaculture 2022, 557, 738332. [Google Scholar] [CrossRef]
  43. Gürüler, H.; Peker, M.; Baysal, Ö. Soft computing model on genetic diversity and pathotype differentiation of pathogens: A novel approach. Electron. J. Biotechnol. 2015, 18, 347–354. [Google Scholar] [CrossRef]
  44. Ornella, L.; Tapia, E. Supervised machine learning and heterotic classification of maize (Zea mays L.) using molecular marker data. Comput. Electron. Agric. 2010, 74, 250–257. [Google Scholar] [CrossRef]
  45. Sousa, I.C.; Nascimento, M.; Silva, G.N.; Nascimento, A.C.; Cruz, C.D.; Almeida, D.P.; Pestana, K.N.; Azevedo, C.F.; Zambolim, L.; Caixeta, E.T. Genomic prediction of leaf rust resistance to Arabica coffee using machine learning algorithms. Sci. Agric. 2020, 78, e20200021. [Google Scholar] [CrossRef]
  46. Torkzaban, B.; Kayvanjoo, A.H.; Ardalan, A.; Mousavi, S.; Mariotti, R.; Baldoni, L.; Ebrahimie, E.; Ebrahimi, M.; Hosseini-Mazinani, M. Machine learning based classification of microsatellite variation: An effective approach for phylogeographic characterization of olive populations. PLoS ONE 2015, 10, e0143465. [Google Scholar] [CrossRef] [PubMed]
  47. Khorramifar, A.; Rasekh, M.; Karami, H.; Malaga-Toboła, U.; Gancarz, M. A machine learning method for classification and identification of potato cultivars based on the reaction of MOS type sensor-array. Sensors 2021, 21, 5836. [Google Scholar] [CrossRef]
  48. Beiki, A.H.; Saboor, S.; Ebrahimi, M. A new avenue for classification and prediction of olive cultivars using supervised and unsupervised algorithms. PLoS ONE 2012, 7, e44164. [Google Scholar] [CrossRef]
  49. Kim, K.J.; Eom, S.H.; Lee, S.P.; Jung, H.S.; Kamoun, S.; Lee, Y.S. A genetic marker associated with the A1 mating type locus in Phytophthora infestans. J. Microbiol. Biotechnol. 2005, 15, 502–509. [Google Scholar]
  50. Brylińska, M.; Sobkowiak, S.; Stefańczyk, E.; Śliwka, J. Potato cultivation system affects population structure of Phytophthora infestans. Fungal Ecol. 2016, 20, 132–143. [Google Scholar] [CrossRef]
  51. Janiszewska, M.; Sobkowiak, S.; Stefańczyk, E.; Śliwka, J. Population structure of Phytophthora infestans from a single location in Poland over a long period of time in context of weather conditions. Microb. Ecol. 2021, 81, 746–757. [Google Scholar] [CrossRef]
  52. Kamvar, Z.N.; Brooks, J.C.; Grünwald, N.J. Novel R tools for analysis of genome-wide population genetic data with emphasis on clonality. Front. Genet. 2015, 6, 208. [Google Scholar] [CrossRef]
  53. Kamvar, Z.N.; Tabima, J.F.; Grünwald, N.J. Poppr: An R package for genetic analysis of populations with clonal, partially clonal, and/or sexual reproduction. PeerJ 2014, 2, e281. [Google Scholar] [CrossRef] [PubMed]
  54. R Core Team. R: A Language and Environment for Statistical Computing; R Foundation for Statistical Computing: Vienna, Austria, 2022. [Google Scholar]
  55. Botalb, A.; Moinuddin, M.; Al-Saggaf, U.M.; Ali, S.S. Contrasting convolutional neural network (CNN) with multi-layer perceptron (MLP) for big data analysis. In Proceedings of the 2018 International Conference on Intelligent and Advanced System, Kuala Lumpur, Malaysia, 13–14 August 2018. [Google Scholar]
  56. Nasien, D.; Yuhaniz, S.S.; Haron, H. Statistical learning theory and support vector machines. In Proceedings of the 2010 Second International Conference on Computer Research and Development, Kuala Lumpur, Malaysia, 7–10 May 2010; pp. 760–764. [Google Scholar]
  57. Jackulin, C.; Murugavalli, S. A comprehensive review on detection of plant disease using machine learning and deep learning approaches. Meas. Sens. 2022, 24, 100441. [Google Scholar] [CrossRef]
  58. Gardner, M.W.; Dorling, S.R. Artificial neural networks (the multilayer perceptron)—A review of applications in the atmospheric sciences. Atmos. Environ. 1998, 32, 2627–2636. [Google Scholar] [CrossRef]
  59. Brownlee, J. Deep Learning for Time Series Forecasting: Predict the Future with MLPs, CNNs and LSTMs in Python; Machine Learning Mastery: New York, NY, USA, 2018. [Google Scholar]
  60. Montesinos-López, O.A.; Montesinos-López, A.; Pérez-Rodríguez, P.; Barrón-López, J.A.; Martini, J.W.; Fajardo-Flores, S.B.; Gaytan-Lugo, L.S.; Santana-Mancilla, P.C.; Crossa, J. A review of deep learning applications for genomic selection. BMC Genom. 2021, 22, 19. [Google Scholar] [CrossRef] [PubMed]
  61. Szűgyi-Reiczigel, Z.; Ladányi, M.; Bisztray, G.D.; Varga, Z.; Bodor-Pesti, P. Morphological Traits Evaluated with Random Forest Method Explains Natural Classification of Grapevine (Vitis vinifera L.) Cultivars. Plants 2022, 11, 3428. [Google Scholar] [CrossRef] [PubMed]
  62. Friedl, M.A.; Brodley, C.E. Decision tree classification of land cover from remotely sensed data. Remote Sens. Environ. 1997, 61, 399–409. [Google Scholar] [CrossRef]
  63. Chauhan, A.S.; Varre, M.S.; Izuora, K.; Trabia, M.B.; Dufek, J.S. Prediction of Diabetes Mellitus Progression Using Supervised Machine Learning. Sensors 2023, 23, 4658. [Google Scholar] [CrossRef]
  64. Hossin, M.; Sulaiman, M.N. A review on evaluation metrics for data classification evaluations. Int. J. Data Min. Knowl. Manag. Process 2015, 5, 1. [Google Scholar]
  65. Wu, N.; Liu, F.; Meng, F.; Li, M.; Zhang, C.; He, Y. Rapid and accurate varieties classification of different crop seeds under sample-limited condition based on hyperspectral imaging and deep transfer learning. Front. Bioeng. Biotechnol. 2021, 9, 696292. [Google Scholar] [CrossRef]
  66. Lourenço, V.M.; Ogutu, J.O.; Rodrigues, R.A.; Posekany, A.; Piepho, H.P. Genomic prediction using machine learning: A comparison of the performance of regularized regression.; ensemble.; instance-based and deep learning methods on synthetic and empirical data. BMC Genom. 2024, 25, 152. [Google Scholar] [CrossRef]
  67. Silva, P.P.; Gaudillo, J.D.; Vilela, J.A.; Roxas-Villanueva, R.M.; Tiangco, B.J.; Domingo, M.R.; Albia, J.R. A machine learning-based SNP-set analysis approach for identifying disease-associated susceptibility loci. Sci. Rep. 2022, 12, 15817. [Google Scholar] [CrossRef]
  68. Yoosefzadeh-Najafabadi, M.; Eskandari, M.; Torabi, S.; Torkamaneh, D.; Tulpan, D.; Rajcan, I. Machine-learning-based genome-wide association studies for uncovering QTL underlying soybean yield and its components. Int. J. Mol. Sci. 2022, 23, 5538. [Google Scholar] [CrossRef]
  69. Benali, S.; Mohamed, B.; Henni, J.E.; Neema, C. Advances of molecular markers application in plant pathology research. Eur. J. Sci. Res. 2011, 50, 110–123. [Google Scholar]
  70. Babarinde, S.; Burlakoti, R.R.; Peters, R.D.; Al-Mughrabi, K.; Novinscak, A.; Sapkota, S.; Prithiviraj, B. Genetic structure and population diversity of Phytophthora infestans strains in Pacific western Canada. Appl. Microbiol. Biotechnol. 2024, 108, 237. [Google Scholar] [CrossRef]
  71. Li, Y.; Cooke, D.E.; Jacobsen, E.; van der Lee, T. Efficient multiplex simple sequence repeat genotyping of the oomycete plant pathogen Phytophthora infestans. J. Microbiol. Methods 2013, 92, 316–322. [Google Scholar] [CrossRef] [PubMed]
  72. Li, Y. Multiplex SSR Analysis of Phytophthora infestans in Different Countries and the Importance for Potato Breeding; Wageningen University and Research: Wageningen, The Netherlands, 2012. [Google Scholar]
  73. Tenzer, I.; degli Ivanissevich, S.; Morgante, M.; Gessler, C. Identification of microsatellite markers and their application to population genetics of Venturia inaequalis. Phytopathology 1999, 89, 748–753. [Google Scholar] [CrossRef] [PubMed]
  74. Varshney, R.K.; Graner, A.; Sorrells, M.E. Genic microsatellite markers in plants: Features and applications. Trends Biotechnol. 2005, 23, 48–55. [Google Scholar] [CrossRef] [PubMed]
  75. Yuen, J.E.; Andersson, B. What is the evidence for sexual reproduction of Phytophthora infestans in Europe? Plant Pathol. 2013, 62, 485–491. [Google Scholar] [CrossRef]
  76. Saville, A.; Ristaino, J.B. Genetic structure and subclonal variation of extant and recent US lineages of Phytophthora infestans. Phytopathology 2019, 109, 1614–1627. [Google Scholar] [CrossRef]
  77. Kiiker, R.; Skrabule, I.; Ronis, A.; Cooke, D.E.; Hansen, J.G.; Williams, I.H.; Mänd, M.; Runno-Paurson, E. Diversity of populations of Phytophthora infestans in relation to patterns of potato crop management in Latvia and Lithuania. Plant Pathol. 2019, 68, 1207–1214. [Google Scholar] [CrossRef]
  78. Vogel, G.; Gore, M.A.; Smart, C.D. Genome-wide association study in New York Phytophthora capsici isolates reveals loci involved in mating type and mefenoxam sensitivity. Phytopathology 2021, 111, 204–216. [Google Scholar] [CrossRef] [PubMed]
  79. Judelson, H.S.; Roberts, S. Multiple loci determining insensitivity to phenylamide fungicides in Phytophthora infestans. Phytopathology 1999, 89, 754–760. [Google Scholar] [CrossRef] [PubMed]
  80. Uddin, S.; Khan, A.; Hossain, M.E.; Moni, M.A. Comparing different supervised machine learning algorithms for disease prediction. BMC Med. Inform. Decis. Mak. 2019, 19, 281. [Google Scholar] [CrossRef] [PubMed]
  81. Misra, S.; Li, H.; He, J. Machine Learning for Subsurface Characterization; Gulf Professional Publishing: Houston, TX, USA, 2019. [Google Scholar]
  82. Meher, P.K.; Begam, S.; Sahu, T.K.; Gupta, A.; Kumar, A.; Kumar, U.; Rao, A.R.; Singh, K.P.; Dhankher, O.P. ASRmiRNA: Abiotic stress-responsive miRNA prediction in plants by using machine learning algorithms with pseudo K-tuple nucleotide compositional features. Int. J. Mol. Sci. 2022, 23, 1612. [Google Scholar] [CrossRef] [PubMed]
  83. Chaitra, N.; Vijaya, P.A.; Deshpande, G. Diagnostic prediction of autism spectrum disorder using complex network measures in a machine learning framework. Biomed. Signal Process. Control 2020, 62, 102099. [Google Scholar] [CrossRef]
  84. Budhlakoti, N.; Kushwaha, A.K.; Rai, A.; Chaturvedi, K.K.; Kumar, A.; Pradhan, A.K.; Kumar, U.; Kumar, R.R.; Juliana, P.; Mishra, D.C.; et al. Genomic selection: A tool for accelerating the efficiency of molecular breeding for development of climate-resilient crops. Front. Genet. 2022, 13, 832153. [Google Scholar] [CrossRef] [PubMed]
  85. Ban, H.J.; Heo, J.Y.; Oh, K.S.; Park, K.J. Identification of type 2 diabetes-associated combination of SNPs using support vector machine. BMC Genet. 2010, 11, 26. [Google Scholar] [CrossRef]
  86. Listgarten, J.; Damaraju, S.; Poulin, B.; Cook, L.; Dufour, J.; Driga, A.; Mackey, J.; Wishart, D.; Greiner, R.; Zanke, B. Predictive models for breast cancer susceptibility from multiple single nucleotide polymorphisms. Clin. Cancer Res. 2004, 10, 2725–2737. [Google Scholar] [CrossRef]
  87. Uhmn, S.; Kim, D.H.; Ko, Y.W.; Cho, S.; Cheong, J.; Kim, J. A study on application of single nucleotide polymorphism and machine learning techniques to diagnosis of chronic hepatitis. Expert Syst. 2009, 26, 60–69. [Google Scholar] [CrossRef]
  88. Yoon, Y.; Song, J.; Hong, S.H.; Kim, J.Q. Analysis of multiple single nucleotide polymorphisms of candidate genes related to coronary heart disease susceptibility by using support vector machines. Clin. Chem. Lab. Med. 2003, 41, 529–534. [Google Scholar] [CrossRef]
Microorganisms 12 00982 g001
Figure 1. Genotype accumulation curve for the P. infestans population for mating type dataset (a), and metalaxyl dataset (b) over 12 loci. The horizontal axis represents the number of loci randomly sampled without replacement up to (n − 1) loci, while the vertical axis shows the number of multilocus genotypes observed. The red dashed line represents 100% of the total observed multilocus genotypes.
Figure 1. Genotype accumulation curve for the P. infestans population for mating type dataset (a), and metalaxyl dataset (b) over 12 loci. The horizontal axis represents the number of loci randomly sampled without replacement up to (n − 1) loci, while the vertical axis shows the number of multilocus genotypes observed. The red dashed line represents 100% of the total observed multilocus genotypes.
Microorganisms 12 00982 g001
Microorganisms 12 00982 g002
Figure 2. Locus and allele count of P. infestans isolates sampled for 12 loci from the Estonia, Pskov region, and Polish P. infestans population.
Figure 2. Locus and allele count of P. infestans isolates sampled for 12 loci from the Estonia, Pskov region, and Polish P. infestans population.
Microorganisms 12 00982 g002
Microorganisms 12 00982 g003
Figure 3. (a) Confusion matrix of training and (b) testing set for mating types (using all marker alleles) for random forest (RF), decision tree classifier, support vector machine (SVM), and artificial neural network (ANN). Coding as follows: 0—A1, 1—A2.
Figure 3. (a) Confusion matrix of training and (b) testing set for mating types (using all marker alleles) for random forest (RF), decision tree classifier, support vector machine (SVM), and artificial neural network (ANN). Coding as follows: 0—A1, 1—A2.
Microorganisms 12 00982 g003
Microorganisms 12 00982 g004
Figure 4. (a) Confusion matrix of training and (b) testing set for metalaxyl response (using all marker alleles) for random forest (RF), decision tree classifier, support vector machine (SVM), and artificial neural network (ANN). NB: Coded; 0: intermediate; 1: resistance; 2: sensitive.
Figure 4. (a) Confusion matrix of training and (b) testing set for metalaxyl response (using all marker alleles) for random forest (RF), decision tree classifier, support vector machine (SVM), and artificial neural network (ANN). NB: Coded; 0: intermediate; 1: resistance; 2: sensitive.
Microorganisms 12 00982 g004
Table 1. Overall prediction metrics for mating type and metalaxyl response based on all SSR markers alleles and common SSR markers alleles.
Table 1. Overall prediction metrics for mating type and metalaxyl response based on all SSR markers alleles and common SSR markers alleles.
All SSR AllelesMating TypeAccuracyPrecisionRecallF1_Score
Random Forest76.720.7260.7470.736
Decision Tree68.250.5700.6680.615
Support Vector Machine70.370.6570.6710.663
Artificial Neural Network37.570.5130.5670.539
Metalaxyl responseAccuracyPrecisionRecallF1_Score
Random Forest75.810.5420.7000.611
Decision Tree68.280.4460.5200.480
Support Vector Machine73.660.4990.6730.573
Artificial Neural Network58.600.3540.3870.370
Mating typeAccuracyPrecisionRecallF1_Score
Common SSR AllelesRandom Forest75.660.7180.7330.726
Decision Tree67.200.5650.6340.598
Support Vector Machine67.720.6190.6370.628
Artificial Neural Network64.550.5590.5850.572
Metalaxyl responseAccuracyPrecisionRecallF1_Score
Random Forest74.190.5380.6520.590
Decision Tree72.040.4730.6700.555
Support Vector Machine73.660.5090.6910.586
Artificial Neural Network63.440.4530.4890.471
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Agho, C.A.; Śliwka, J.; Nassar, H.; Niinemets, Ü.; Runno-Paurson, E. Machine Learning-Based Identification of Mating Type and Metalaxyl Response in Phytophthora infestans Using SSR Markers. Microorganisms 2024, 12, 982. https://doi.org/10.3390/microorganisms12050982

AMA Style

Agho CA, Śliwka J, Nassar H, Niinemets Ü, Runno-Paurson E. Machine Learning-Based Identification of Mating Type and Metalaxyl Response in Phytophthora infestans Using SSR Markers. Microorganisms. 2024; 12(5):982. https://doi.org/10.3390/microorganisms12050982

Chicago/Turabian Style

Agho, Collins A., Jadwiga Śliwka, Helina Nassar, Ülo Niinemets, and Eve Runno-Paurson. 2024. "Machine Learning-Based Identification of Mating Type and Metalaxyl Response in Phytophthora infestans Using SSR Markers" Microorganisms 12, no. 5: 982. https://doi.org/10.3390/microorganisms12050982

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop