A Practical Approach to Genome Assembly and Annotation of Basidiomycota Using the Example of Armillaria

Abstract

Technological advancements in genome sequencing, assembly and annotation platforms and algorithms that resulted in several genomic studies have created an opportunity to further our understanding of the biology of phytopathogens, including Armillaria species. Most Armillaria species are facultative necrotrophs that cause root- and stem-rot, usually on woody plants, significantly impacting agriculture and forestry worldwide. Genome sequencing, assembly and annotation in terms of samples used and methods applied in Armillaria genome projects are evaluated in this review. Infographic guidelines and a database of resources to facilitate future Armillaria genome projects were developed. Knowledge gained from genomic studies of Armillaria species is summarized and prospects for further research are provided. This guide can be applied to other diploid and dikaryotic fungal genomics.

TWEETABLE ABSTRACT

We reviewed methods used in Armillaria (Fungi) genome projects. Infographic guidelines and a resource database were developed to facilitate future genome projects. This guide can be applied to other diploid and dikaryotic genomics.

Graphical Abstract

Keywords::

• bioinformatics
• DNA extraction
• evolution
• genomics
• metabolomics
• next-generation sequencing
• phylogenomics
• phytopathogens
• proteomics
• transcriptomics

Armillaria includes about 70 known species [1] that are distributed worldwide, as reviewed by Baumgartner et al. and Coetzee et al. [2,3]. Two species, Armillaria ectypa and Armillaria tabescens, that were previously placed in Armillaria, now reside in Desarmillaria based on absence of annulus at maturity and an evolutionary trajectory separated from the rest of the Armillaria species [4]. Armillaria spp. are ubiquitous, have a broad host range, parasitize both herbaceous and woody plants and survive under diverse environmental conditions, as reviewed by Heinzelmann et al. [5]. The ubiquity of Armillaria spp. has been attributed to their life cycles, lifestyles and other biotic factors. We refer the reader to previous reviews for further information regarding the ecology, life cycles, lifestyles, species boundaries and geographical distribution and other properties of Armillaria spp [2,3,5].

Armillaria spp. are of significant economic and agroecological importance due to various characteristics they exhibit. When they are parasitic, they cause massive devastation in horticulture and agriculture, as well as in natural and managed plantation forests [6–9]. However, Armillaria spp. enhance the forest ecosystem when they act as saprophytes [10]. The forest ecosystem is enriched when saprobic microorganisms including Armillaria spp. degrade plant biomass with their extensive enzyme systems, resulting in significant contribution in releasing sequestered organic carbon into the carbon cycle [11,12]. Moreover, like other saprobic Basidiomycota species, some species in the genus are also edible. Hence there is interest in domesticating species such as Armillaria mellea for culinary purposes, as reviewed by Ren et al. [13]. Additionally, the enzymes, secondary metabolites and other compounds produced by Armillaria spp. have the potential to be used in various industries such as the pharmaceutical industry. Therefore research on Armillaria spp. has focused on understanding the growth mechanisms in terms of in situ and in vitro mycelia, rhizomorph and fruit body growth, as well as pharmacological and other properties [13,14]. Other research on these fungi has focused on mechanisms which drive their pathogenicity and can be exploited and/or harnessed to halt their growth (in the case of pathogenicity of socioeconomically important plants) or enhance their production of enzymes and other compounds of interest for biotechnological applications [13,14]. Despite the research milestones made to better understand and limit or utilize these mushroom-forming fungi, there are still large knowledge gaps. Closing of these gaps is relevant as the current management practices have proven to be mostly inefficient in managing Armillaria root-rot disease [5,13].

Several factors facilitate survival and pathogenicity and/or virulence of Armillaria spp. To understand these mechanisms and to detect possible applications, genomics-based research has been conducted by various research groups on Armillaria spp. [11,15–22]. These studies have given some insights into various characteristics of Armillaria spp., mostly from samples collected from countries in the northern hemisphere. In the present review, the extent of genomic studies on Armillaria species is considered, with a focus on the diversity of species and strains for which genomes have been assembled and annotated. The steps, techniques and tools used for genome extraction, sequencing, assembly, evaluation and annotation are discussed. Infographic guidelines for future Armillaria genome projects with some relevant resources have also been provided. Knowledge gained from these genome projects is also highlighted. We further suggest potential areas for future Armillaria genome projects as well as potential comparative genomic and multi-omic studies. Although this review centers around genomic studies of Armillaria spp., the methods and information gathered from these studies can be extended to genome research on other fungi.

Status of genome sequencing & assembly of Armillaria spp.

There are presently (March 2023) 11 published whole genome sequences of Armillaria spp. in the National Center for Biotechnology Information’s (NCBI) Genome [23] and US Department of Energy’s Joint Genome Institute (JGI) fungal genomics resource MycoCosm databases (Table 1) [24]. Other genomes have been submitted to the JGI MycoCosm database but have not yet been published in journals and require permissions from the principal investigators for use. This review will therefore focus on the published genomes of Armillaria spp. only and the knowledge gained from their genomic studies. For concision, we refer the reader to links and references provided for all the tools discussed in this review in Supplementary Table 1.

Table 1. Methods and tools used in sequencing and assembly of published genomes of Armillaria species to date (March 2023).

Species [strain or isolate]	Country of origin (lifestyle)^†	DNA extraction methods/kits	Sequencing technologies (pre-assembly reads correction)	Genome assembly tools	Post-assembly processing	Genome completeness evaluation	NCBI accession number or genome link	Ref.
Armillaria altimontana (NABS X) [837–10]	USA (unknown)	MoBio (Qiagen) DNeasy PowerMax Soil Kit	PacBio SMRT cells (NA)	Pacific Biosciences SMRT Portal software	NA	QUAST, BUSCO 2.0b2 with Basidiomycota dataset	JAIWYR000000000.1	[19]
Armillaria borealis [AB13-TR4-IP16]	Eastern Siberia (FN, weakly pathogenic, opportunistic)	Modified version of the hot-CTAB extraction at 65°C with chloroform washing	Illumina MiSeq (NA)	SPAdes v. 3.13.0 genome assembler	NA	BUSCO and RS BridgeMapper protocol	JAAGUC000000000.1	[21]
Armillaria cepistipes [B5]	Italy (FN, weakly pathogenic)	PowerMax MOBIO DNA isolation kit	PacBio RS II platform; Illumina HiSeq 2000 (NA)	HGAP3 workflow of the SMRT Analysis suite v. 2.3	PBJelly2, FinisherSC, RS resequencing protocol and Pilon	CEGMA and BUSCO v. 3.0.1 with the lineage-specific profile library basidiomycota_odb9 and RS resequencing protocol	FTRY00000000.1	[16]
Armillaria fuscipes [CMW 2740]	South Africa (FN, particularly pathogenic to exotic species)	DNeasy Plant Mini Kit (Qiagen)	Illumina HiSeq; PacBio sequencing with 20 SMRT cells (NA)	Celera Assembler pipeline, CLC Genomics Workbench v. 5.5.1 and MIRA	Velvet optimizer, Graph Accordance Assembly program and SSPACE	CEGMA and BUSCO using gene models from Laccaria bicolor	LWUH00000000.1	[18]
Armillaria gallica [012 m]	China (FN, weakly or secondarily pathogenic)	DNeasy plant kit (Qiagen)	PacBio Sequel; Illumina HiSeq (LoRDEC v.0.6, Trimmomatic v.0.36)	ALLPATHS-LG V52488 and MECAT2	Purge Haplotigs, FinisherSC v. 2.1 and NextPolish v. 1.0.21	BUSCO with the lineage-specific profile library basidiomycota_odb9	VTST00000000.1	[20]
Armillaria gallica [Ar21-2]	USA (FN, weakly or secondarily pathogenic)	PowerMax MOBIO DNA isolation kit	Illumina MiSeq (NA)	HGAP3 workflow of the SMRT Analysis suite v. 2.3	PBJelly2, FinisherSC, RS Resequencing protocol and Pilon	CEGMA and BUSCO v. 3.0.1 with the lineage-specific profile library basidiomycota_odb9 and RS BridgeMapper protocol	NKEW00000000.1	[16]
Armillaria mellea [DSM 3731]	France (FN, highly pathogenic)	ZR Fungal/Bacterial DNA Kit using a modified protocol	Illumina GAII (NA)	Velvet v. 0.7.53	NA	NA	https://genome.jgi.doe.gov/portal/Armme1_1/Armme1_1.download.html	[17]
Armillaria ostoyae [C18/9]	Switzerland (pathogenic)	PowerMax MOBIO DNA isolation kit	PacBio RS II platform; Illumina HiSeq 2000 (NA)	HGAP3 workflow of the SMRT Analysis suite v. 2.3	PBJelly2, FinisherSC, RS Resequencing protocol and Pilon	CEGMA and BUSCO v. 3.0.1 with the lineage specific profile library basidiomycota_odb9 and RS BridgeMapper protocol	FUEG00000000.1	[16]
Armillaria solidipes [C28-4]	USA (FN, highly pathogenic)	PowerMax MOBIO DNA isolation kit	Illumina HiSeq 2000 (NA)	HGAP3 workflow of the SMRT Analysis suite v. 2.3	PBJelly2, FinisherSC, RS Resequencing protocol and Pilon	CEGMA and BUSCO v. 3.0.1 with the lineage specific profile library basidiomycota_odb9 and RS BridgeMapper protocol	NKHM00000000.1	[16]
Armillaria solidipes (form A. ostoyae) [ID001]	USA (FN, highly pathogenic)	MOBio DNeasy PowerMax Soil Kit	PacBio SMRT cells (NA)	Pacific Biosciences SMRT Portal software	NA	QUAST and BUSCO 2.0b2 with Basidiomycota dataset	JAIWYQ000000000.1	[19]
Armillaria African Clade B sp. [CMW4456]	Zimbabwe (unknown)	Qiagen DNEasy Plant Pro Kit (50)	PacBio HiFi; Illumina HiSeq (Trimmomatic)	CLC Genomics Workbench v. 22.0.1	Pilon v. 1.23	QUAST v 5.0.2 and BUSCO v. 5.3.2, using the agaricales_odb10 lineage dataset in the Galaxy platform	JANDKJ000000000.1	[22]

† Lifestyle information for A. ostoyae was obtained from [25,26]. Lifestyle information for other species was summarized from [4].

CTAB: Cetyltrimethylammonium bromide; FN: Facultative necrotroph; NA: Not available; NCBI: National Center for Biotechnology Information.

Starting materials & DNA extraction methods used in Armillaria genome projects

The starting material can influence the quantity and quality of DNA extracted for downstream processes. The quality of DNA extracted, in terms of chemical purity, structural integrity and size of the DNA, influences the choice of sequencing technology that can be employed in a genome project. These parameters also affect the quality of obtained data for downstream processes, including genome assembly [27]. Armillaria genome projects have typically used fresh mycelia, mostly obtained as liquid cultures in different media, as the starting materials (Table 1) [16–22]. For the Armillaria borealis strain AB13-TR4-IP16 genome project, fresh mycelium isolated from under the bark of infested Abies sibirica stems 50 cm above the soil surface was used as starting material [21].

Relatively low-molecular-weight DNA will be sufficient for sequencing using second-generation sequencing (SGS) technologies (also known as short reads; e.g., Illumina, 454, SOLiD, Ion Torrent). High-molecular-weight DNA in sufficient quantity and quality is required for third-generation sequencing (TGS, also known as long reads; e.g., Pacific Biosciences [PacBio], Oxford Nanopore Technologies). Typically, the minimum amounts of DNA required for Illumina, PacBio, Oxford Nanopore and BioNano are approximately 3 ng, 20 μg, 1 μg and >200 ng, respectively [28]. High-molecular-weight DNA is typically obtained from fresh material [27]. Other considerations for DNA quality requirements for de novo sequencing have been discussed by Del Angel et al. [27].

DNA extraction for genome projects can be achieved using various protocols or DNA extraction kits. The cetyltrimethylammonium bromide method and modifications of it have been used to extract DNA from fungal and plant samples, which characteristically contain high concentrations of impurities (RNA, polysaccharides, proteoglycans, proteins, secondary metabolites, polyphenols and other contaminants) [29–33]. Akulova et al. used a modified cetyltrimethylammonium bromide method for extracting DNA from A. borealis strain AB13-TR4-IP16 [21]. Commercial DNA extraction kits have also been used in Armillaria genome projects, either according to the manufacturer’s instructions or with some modifications [16–20,22]. These kits produce high-quality DNA extracts but may differ in their efficiency in terms of the quality and quantity of DNA extracted in relation to the specific organism [34,35]. Relatively high concentrations of some compounds considered to be impurities in DNA extracted for genome sequencing – including polysaccharides, proteins, secondary metabolites and polyphenols – have been reported in liquid cultures and fruiting bodies of A. mellea and Armillaria gallica grown under different conditions and extracted with different solvents [36–39]. For this reason, DNA extraction protocols or kits adopted for use in Armillaria genome projects should aim at reducing or eliminating these impurities to obtain high chemical purity, while maintaining the structural integrity and size of the DNA.

Choice of genome sequencing technologies

Armillaria genome projects have employed both SGS (i.e., Illumina HiSeq and Illumina MiSeq) and TGS (i.e., PacBio RS and PacBio SMRT) technologies (Table 1), similar to other fungal genome sequencing projects. These technologies have their respective strengths and limitations, as reviewed by several authors [27,40–42]. Although the SGS technologies are relatively cheaper, the generation of long contiguous DNA sequence reads during genome assembly is often impaired due to short read lengths, as well as the presence of repeat elements and polymorphisms [40,41]. Genome assemblies from reads obtained from SGS technologies also result in the loss of as much as 99.1% of validated duplicated sequences [38]. These challenges are largely circumvented by TGS, which provides long reads for better representation of the genome being assembled, although high rates (5–20%) of sequencing error of raw reads are obtained with this technology [40–42]. However, the technologies are being improved to reduce these error rates.

Five of the Armillaria genome projects utilized both SGS and TGS technologies for hybrid de novo genome assemblies (Table 1). These were Armillaria cepistipes strain B5, Armillaria fuscipes strain CMW 2740, A. gallica strain 012 m, Armillaria ostoyae strain C18/9 and Armillaria African Clade B sp. strain CMW4456. Hybrid approaches were followed to minimize the effects of individual use of SGS and TGS technologies on the quality and contiguity of the assembled genomes. Four of these genome projects used the Illumina reads for error correction or polishing after the PacBio read assembly [16,18,20], while the reverse approach was used for assembling the A. fuscipes strain CMW 2740 genome [22]. This hybrid genome assembly method has been used in other fungal genome projects such as the Fusarium equiseti genome project [43] and is the most frequently used combination of techniques in bacterial RefSeq genome projects [44]. Another combination of genome assembly techniques is the use of Illumina and MinION™ reads [44,45]. MinION is another Oxford Nanopore TGS technology [46]. This combination has not yet been used in Armillaria genome projects and might be useful for assembling good-quality genomes in future.

Raw reads quality control

Quality control (QC) pre-assembly is highly important as this can impact the quality of the final assembled genome. Methods or bioinformatic tools used for the QC steps in the published Armillaria genome projects were usually not reported. Nonetheless, various tools available for this purpose include FastQC, FastQE, Falco (FastQC Alternative Code), NanoPlot, PycoQC, LongQC and MultiQC (Figure 1 & Supplementary Table 1). These tools assess quality metrics including base or sequence quality of the reads, presence of adapter sequences and GC distribution. The choice of the tools used for raw-reads QC is usually informed by the sequencing technology used and the number of genomes being assembled. Additionally, raw-reads QC is sometimes integrated into genome assembly pipelines. As such, the decision of whether to perform a separate QC step should consider the genome assembly tool or pipeline to be used in the next steps.

Figure 1. Workflow, technological tools and decision-making considerations for genome sequencing and assembly.

Pre-assembly reads correction

Pre-assembly correction of raw reads is optional in genome projects because this step depends on the quality of the raw reads obtained, the purpose of the genome project and computational time, as well as the choice of genome assembly tool or workflow in the next step [27]. Some genome assembly tools or workflows incorporate reads correction; hence a separate reads correction is not needed in this case.

Tools such as LoRDEC and Trimmomatic have been used for reads correction in the Armillaria genome projects (Table 1). LoRDEC is mostly intended to correct PacBio reads, using Illumina reads as reference. Trimmomatic performs trimming of adapters and other Illumina-specific sequences, as well as quality and length trimming for Illumina paired-end and single-end reads. The order of the steps used during reads correction with Trimmomatic is important and depends on the quality of the raw sequence data. Other available tools for reads profiling and correction include the Trim Reads function in CLC Genomics Workbench (Qiagen, Aarhus, Denmark), Cutadapt, Filtlong and Porechop (Figure 1 & Supplementary Table 1). These tools also perform adapter trimming, quality trimming and length trimming for reads obtained from SGS, TGS or both. The choice of which tool(s) to use depends on the sequencing technology used, the genome assembly pipeline and other considerations, some of which are outlined in Figure 1.

Genome assembly tools, pipelines or workflows

De novo genome assembly using various genome assembly tools has been used as opposed to reference-based assembly in Armillaria genome projects (Table 1). Genome assembly tools that were used include CLC Genomics Workbench v. 22.0.1, SPAdes, Velvet, MIRA, ALLPATHS-LG, MECAT2, PacBio SMRT Portal software and the now obsolete Celera Assembly Pipeline (Table 1). These tools have their respective strengths and limitations. For the specific functions and considerations for selecting and working with specific genome assembly tools, pipelines or workflows, see the links, references and workflow provided in Figure 1 & Supplementary Table 1.

De novo genome assembly is impacted by the computational methods used due to constant technological advancement and specific limitations of respective genome assemblers. The nature and quantity of sequence data, which rely on the features of the genome being assembled, also impact de novo genome assembly [47,48]. Differences in various genome assemblers – in terms of memory consumption, total assembling time and accuracy based on genome quality matrices (contig number, N50 values and genome fractions) using prokaryotic and eukaryotic Illumina paired- and single-end reads – have been evaluated by Khan et al. [49]. The differences of the various assemblers and the unique biological characteristics of the genome being assembled require that correct assembler versions should be selected, and various assemblers and assembly parameters should be evaluated in genome projects. Outputs of the various assemblies should be compared in each genome project to determine the optimum assembly for the specific genome. Nevertheless, knowing when to stop while chasing perfection in a genome project is vital. It is essential to consider the complexity of the genome to be assembled (i.e., genome size, amount and distribution of repeat sequences, heterozygosity, ploidy and homogeneity of GC content), the research questions and available resources (e.g., availability of a reference genome, project budget, time and computational constraints) in the decision-making process [27,50].

Post-assembly processing

Post-assembly processing involves sequence error correction, polishing and haplotype phasing. Assembly errors are introduced especially when long reads are assembled. Hence post-assembly error correction and polishing is highly recommended to improve the quality of the overall local base accuracy and resolve some misassemblies of the assembled genome. This can be achieved by polishing the assembly with short or long reads [27]. Tools which have been used in Armillaria genome projects for this purpose include NextPolish, Pilon, FinisherSC, PBJelly2, RS Resequencing application (SMRT Analysis suite) and VelvetOptimiser (Table 1).

Haplotype phasing is the identification of alleles that are co-located on the same chromosome. Haplotype phasing and removal of the identified heterozygous overlaps in assembled genomes is relevant to diploid or polyploid genomes and in genomes which have regions of high heterozygosity [51]. This is because most algorithms in diploid-aware assemblers (e.g., Canu) will assemble these regions as separate contigs, rather than the expected single haplotype-fused contig, once a pair of allelic sequences exceeds a certain threshold of nucleotide diversity [52]. Armillaria spp. usually exist as prolonged diploid vegetative states [5]; hence most sequenced genomes are diploid and warrant removal of heterozygous overlaps in assembled Armillaria genomes. This can be achieved by utilizing standalone tools such as Purge Haplotigs, as demonstrated in the A. gallica strain 012 m genome project (Table 1) [20]. Purge Haplotigs automates the reassignment of allelic contigs and assists with manual curation of TGS-based genome assemblies specifically [52]. As outlined in Figure 1 & Supplementary Table 1, other tools are available for haplotype phasing and differ in how they function (see provided tool websites and/or references).

Genome quality evaluation

The quality of an assembled genome must be evaluated after genome assembly. This evaluation should be conducted in terms of genome completeness, the amount of misassemblies and the presence of contaminants. Metrics such as N50 sizes and L50 are indicative of the level of contiguity (high N50 and L50 values) of the genome assembly [50,53]. N50 sizes are the contig/scaffold size above which half the genome is represented, while L50 is the length in bases of the shortest contig for which 50% of the genome can be contained within contigs of that size or larger. These two metrics are computed based on the lengths of assembly scaffolds or contigs [50].

Various tools can be used for genome assembly evaluation. Armillaria genome projects have used tools such as BUSCO (‘Benchmarking Universal Single-Copy Orthologs’), QUAST (‘QUality ASsessment Tool’), RS BridgeMapper protocol (SMRT Analysis Suite) and the now defunct CEGMA (‘Core Eukaryotic Genes Mapping Approach’) to evaluate the completeness and quality of the respective genome assemblies (Table 1). BUSCO analyses of Armillaria genome projects have utilized different versions of BUSCO (5.3.2 and 3.0.1) and different lineage-specific datasets (agaricales_odb10 lineage and basidiomycota_odb9 datasets) (Table 1). The results generated by BUSCO depend on the lineage selected and version of database of the lineage used for evaluation of the sequenced organism, as well as the version of the tool itself. For example, the basidiomycota_odb9 dataset contained 1335 BUSCO groups, while the basidiomycota_odb10 dataset contains 1764 BUSCO groups. Moreover, the agaricales_odb10 and eukaryota_odb10 datasets contain 3870 and 255 BUSCO groups, respectively. BUSCO v. 5.4.4 is the current stable version [54]. Consequently, care should be taken when comparing genome completeness based on BUSCO analyses from different genome projects.

In addition to evaluating genome completeness and correctness, genomes should be evaluated for absence of contaminants (i.e., nontarget sequences). Genome contamination can occur and be detected at various stages of a genome project [55]. These contaminants can be detected and removed with various tools which function by using various parameters including detection of a highly skewed (i.e., extremely low or high) GC content; presence of an uncharacteristically high percentage of duplicated complete BUSCO genes; and results of BLAST searches of sections of the genome in the NCBI database [55]. BlobTools and Anvi'o have been used to detect and remove genome contamination from an assembled fungal genome [56]. Aside from BUSCO, none of the indicated tools were used in the Armillaria genome projects as genome contaminants may have been absent in the respective assemblies.

In summary, genome quality evaluation should be conducted in all genome projects. Genomes that seemingly have unsatisfactory completeness and correctness metrics can be reassembled with other tools or workflows with altered assembly parameters, and the best assembly in terms of these metrics selected as the final assembly. Minimal contamination can be resolved with tools such as the ones indicated in the review by Cornet and Baurain [55], some of which have been outlined in the present review (Figure 1 & Supplementary Table 1). The purpose for the genome project can be used to select the best assembly in genome assembly quality evaluation (e.g., selecting between the assembly with the highest contiguity or the assembly with the highest percentage of complete genes). This is important as assembly quality is not well correlated with assembly contiguity [50]. Additionally, it is apparent that genome quality evaluation technologies, in terms of both computation and datasets used, are being rapidly developed and/or enhanced. Hence research teams embarking on genome projects should keep abreast of the state-of-the-art tools for genome assembly and quality evaluation. Tools used for genome quality evaluation in Armillaria genome projects and other available tools compute different metrics which can be accessed via the information provided in Supplementary Table 1. In the case of very low genome completeness and correctness as well as high contamination, it would be more prudent to sequence and assemble a new genome.

Transposable elements & genome annotation

Genome annotation is required to attach biologically meaningful information to assembled genomes, which are otherwise of no significant value to biologists [27]. This process consists of annotation of groups of mobile genetic elements called transposable elements (TEs) as well as gene prediction through ab initio approaches with or without the addition of wet-bench experimental data [27]. Wet-bench experimental data for genome annotation include short-reads sequenced expressed sequence tags, RNA sequencing or even long-read sequenced transcripts [27]. Genome annotation also involves describing the structures and functions of the products of the predicted genes. The reader is referred to two book chapters by De Sá et al. [57] and Christoffels and Heusden [58] as well as the review articles by Yandell and Ence [53] and Del Angel et al. [27] for detailed steps used for genome annotation. As a summary, the methods and depth of Armillaria genome annotation have been outlined in Figure 2 and are described in this section.

Figure 2. Steps and tools used in Armillaria genome annotation.

Levels 1, 2 and 3 are based on the approaches to genome annotation relative to time, effort and evidence used and the resultant increase in annotation accuracy, as outlined by Yandell and Ence [53]. Level 2 can be performed with the ab initio prediction tools indicated in level 3 (point B under ‘Tools, Databases, Pipelines’).

ACB: African Clade B; NA: Not applicable.

Gene and functional annotation of assembled genomes is preceded by identification and masking of RNA genes, detection and annotation of TEs and other repetitive elements and detection of low-complexity DNA regions. TEs include long terminal repeats (LTRs), and their annotation is considered a major task in any genome project [27,59]. RNA gene identification and masking, and detection and annotation of TEs, were carried out in all the Armillaria genome projects excluding the Armillaria African Clade B sp. (CMW4456) genome project using the tools indicated in Figure 2. Tools employed in Armillaria genome projects for identification and masking of RNA genes in the assembled genomes include RNAmmer v. 1.2, Rnnotator, SortMeRNA and tRNAscan-SE [16,19–21]. De novo detection and annotation of TEs, other repetitive elements and low-complexity DNA regions have been achieved with tools including BLASTER, Grouper, LTRHarvest, PASTEC, Piler, Recon, RepBase, RepeatMasker, RepeatModeler, REPET and TEclass [16,18–21]. Although not clearly indicated in the Armillaria African Clade B sp. (CMW4456) genome project, the AUGUSTUS annotation tool on the Galaxy platform has an option called Softmasking, which was used in the Armillaria African Clade B sp. (CMW4456) project. Repeated regions in the genome assembly are detected as lowercase letters when this option is enabled [60].

Gene and functional annotation of Armillaria genomes has either been restricted to single ab initio gene prediction only or has involved full-scale annotation with or without wet-bench experimental evidence and optional manual curation of gene models (Figure 2). For prediction and annotation of genes and proteins, tools including AUGUSTUS, Blast2GO, dbCAN2 server, Exonerate, Fgenesh, GETA, PHI base and SNAP have been employed [16–22]. Some of these tools have been utilized in pipelines such as BRAKER2 and Maker [16].

Functional annotation of some of the Armillaria genomes has incorporated the use of RNA-Seq data [16,20,21]. Annotation with wet-bench experimental data involves transcript assembly using RNA-Seq data. This has been achieved with Trinity and Rnnotator in the Armillaria genomes annotated by Sipos et al. [16] and with SortMeRNA v. 2.1 in the genome annotation reported by Akulova et al. [21]. The tool used for RNA-Seq data assembly was not specified in the genome project by Zhan et al. [20]. Following the use of multiple ab initio predictors, genome annotation in the genome projects reported by Sipos et al. [16] involved manual curation of the predicted gene models using tools or pipelines such as Gbrowse, FunyBASE, PASTEC and PEDANT systems. The authors used the JGI Annotation Pipeline for annotation of the A. gallica strain Ar21-2 and Armillaria solidipes strain C28-4 genomes.

Further genome annotation has been conducted in Armillaria genome projects to gain more insights about these fungi. For this purpose, tools such as BLASTp, GeneMark-ES and GeneMark-ET have been used for gene and protein ortholog finding [16,19,21]. Motifs and domains of proteins have been detected with two versions of InterProScan in the Armillaria altimontana (NABS X) strain 837–10, A. gallica strain 012 m and A. solidipes strain ID001 genome projects [19,20]. BLASTp, SecretomeP server and SignalP were used for putative gene function predictions based on databases and/or protein evidence [16,17,19,21]. Only the A. altimontana (NABS X) strain 837–10 and A. solidipes strain ID001 genome projects predicted subcellular localizations of the genes and annotated the secondary metabolite biosynthesis gene clusters [19]. These were achieved with Deeploc 1.0 and antiSMASH, respectively. Evaluation of the annotations was achieved using BUSCO in the genome projects reported by Sipos et al. [16] and Caballero et al. [19].

It is apparent that genome annotation in Armillaria genome projects has ranged from minimal annotation to in-depth annotation. The investments in terms of time, effort and evidence (with a consequence on the project budget) needed for genome annotation increase with increasing accuracy of the annotation, as explained by Yandell and Ence [53]. Therefore it is relevant to consider the project budget and other resources, including time and computational constraints, as well as the research questions, to determine whether to perform only in silico gene and functional annotation or to include wet-bench experimental evidence with extensive annotations. Further considerations for genome annotation have been discussed by Christoffels and van Heusden [58]. Annotations of all the Armillaria genomes have been used to gain some knowledge about the biology of these fungi. Some of these are discussed in the next section.

Knowledge gained from Armillaria genomics

Armillaria genome projects have provided draft and complete genome assemblies [15–22]. Standardized QUAST and BUSCO analyses of the presently published Armillaria genomes were conducted for accurate comparison in this review (Table 2). This comparison shows that the Armillaria genomes ranged between approximately 50.71 and 87.31 Mbp. These genome sizes are comparable among the Armillaria species and are within the reported genome sizes (~28–175 Mb) of other Agaricales genomes [61]. Even at the draft stage, Armillaria genomes are highly valuable resources with sufficiently high contiguity (N50 values >5000 bp) and completeness (>98% complete BUSCOs) and thus can be used for some comparative genomics and other studies.

Table 2. QUAST and BUSCO evaluation of published Armillaria genomes.

Assembled isolate	QUAST evaluation metrics^‡						Complete BUSCOs^§ (%, n = 3870)
	Contigs (n)	Largest contig (bp)	Total length (bp)	GC (%)	N50 (bp)	Ns per 100 kbp (n)
Armillaria altimontana (NABS X) 837–10	100	5,843,527	73,739,702	47.77	1,930,169	14	98.8 (98.2, 0.6)
Armillaria borealis AB13-TR4-IP16]	10,402	2,136,877	56,994,024	47.76	75,348	5501	97.9 (97.4, 0.5)
Armillaria cepistipes B5	182	6,523,365	75,519,607	47.59	2,288,895	0	98.9 (97.8, 1.1)
Armillaria fuscipes CMW 2740	17,117	157,180	50,706,099	47.68	6152	982	92.6 (88.9, 3.7)
Armillaria gallica 012 m	63	6,431,929	87,305,441	47.38	2,159,699	0	98.7 (95.9, 2.8)
A. gallica Ar21-2	319	4,779,317	85,336,812	47.35	1,035,263	8165	98.6 (97.9, 0.7)
Armillaria mellea DSM 3731^†	9,867	639,705	75,786,598	46.99	26,609	16,520	97.9 (97.5, 0.4)
Armillaria ostoyae C18/9	106	6,405,655	60,106,801	48.33	2,283,935	0	98.8 (98.2, 0.6)
Armillaria solidipes C28-4	229	3,399,694	58,009,494	48.35	715,667	3906	98.9 (98.3, 0.6)
A. solidipes (form A. ostoyae) ID001	72	4,463,803	55,735,298	48.26	2,424,439	5	98.7 (98.2, 0.5)
Armillaria African Clade B sp. CMW4456	840	1,463,441	54,954,005	46.53	128,967	0	98.2 (97.0, 1.2)

Unless otherwise stated, the table was created with analysis of genome statistics of Fasta sequences of published genomes at the National Center for Biotechnology Information Genome [62].

† Downloaded from JGI MycoCosm.

‡ QUAST analyses conducted in Galaxy using the genome assembly mode, and ‘Fungus use of GeneMark-ES for gene finding, Barrnap for ribosomal RNA genes prediction’.

§ BUSCO analyses performed using BUSCO v.5.3.2 in genome mode using AUGUSTUS gene predictor with agaricales_odb10 lineage dataset. BUSCO data are presented as total complete BUSCOs (single-copy BUSCOs, duplicated BUSCOs). There were <5% fragmented and missing BUSCOs each in all the genomes.

BUSCO: Benchmarking universal single-copy ortholog; JGI: Joint Genome Institute.

Some of the Armillaria genomes have been extensively annotated. TE annotation of Armillaria genomes by Sipos et al. revealed considerable variation in TEs among the different genomes and suggested that unlike in other plant pathogens, TEs do not account for genome expansion in Armillaria [16]. Genome expansion of rust genomes, including that of Austropuccinia psidii which has an estimated haploid genome size of just over 1 Gb, have been attributed to TEs [63,64]. LTRs were the most highly represented TE family in the genome of A. borealis strain AB13-TR4-IP16, with Gypsy (32) being the most abundant LTR type followed by Copia (5) [21]. Other repetitive elements reported in this genome included long interspersed nuclear element retrotransposons and retrotransposons with a tyrosine recombinase. Similarly, the most abundant LTR type in the genome of A. gallica strain 012 m was Gypsy (2062) followed by Copia (321) [20]. This trend has been reported in other Basidiomycota genomes [65]. Although the functions of these TEs have not yet been identified in Armillaria, these mobile genetic elements have been speculated or shown to have various functions in the genomes of other organisms. These functions include acting as mutators, contributing to genome restructuring and participation in the evolutionary ‘arms race’ between hosts and pathogens [66]. Nevertheless, the contribution of mobile genetic elements to size variation in the mitochondrial genomes (mitogenomes) of A. borealis, A. gallica, Armillaria sinapina and A. solidipes was discovered through comparative mitogenomics [67]. It is relevant to conduct further studies on the functions of TEs in the genomes of Armillaria spp. to gain a better understanding of the effects of TEs on the evolution, ecology and other characteristics of these fungi.

The number of protein-coding genes identified in all the genomes ranged from 13,600 in Armillaria African Clade B sp. (CMW4456) to 26,261 in A. gallica 012 m [15–22]. The predicted gene contents of the Armillaria genomes are within the range reported by Ruiz-Duenas et al. [61] for other Agaricales genomes (5000–29,000 genes). Although the effect of gene contents (in terms of number of genes) and genome size have not been evaluated in Armillaria, Ruiz-Duenas et al. reported that these properties do not significantly correlate to the lifestyles exhibited by the Agaricales [61].

Information gained from Armillaria genomics has contributed to the body of knowledge regarding their genome evolution and architecture as well as their biology. For instance, Sipos et al. reported that there is a significant genome expansion in Armillaria spp. compared with other related fungi, with a consequent effect on various properties, including gene family expansion of pathogenicity-related genes [16]. There are also records of many relatively small differences in gene contents of genomes of Armillaria spp. which potentially contribute to differences in their ecological lifestyles [11,15,16,19,68]. All these records point to genome plasticity of Armillaria spp. which can mediate adaptability to changing environments and resistance to, or evasion of, host defense mechanisms. These changes can occur through gene inactivation, altered gene sequence or structure and the birth of new genes, which often encode proteins involved in host–pathogen interactions, as reviewed by Raffaele and Kamoun [69].

The availability of genome sequences and consequent genome annotations has allowed for phylogenomic studies on Armillaria spp. In this way, the position of Armillaria in the Physalacriaceae and existence of Guyanagaster and Cylindrobasidium as the closest relatives were confirmed using 835 conserved single-copy genes (188,895 amino acid sites) [16]. Furthermore, the age of pathogenic Armillaria spp. was estimated at 21 million years and their divergence from Guyanagaster is reckoned to be 42 million years [16]. Phylogenomics, by virtue of using large numbers of orthologous genes, provides an advantage over traditional phylogenetic methods that employ few loci for obtaining well-supported and accurate species delimitation [70]. This is explained by the fact that individual gene genealogies may differ from each other and from the organismal phylogeny and therefore affect the phylogeny obtained when the analysis is based on few genes [70]. However, the use of more genes or loci is not always better in phylogenomics. This is because the possibility of obtaining highly supported but incorrect phylogenetic results, that are due to inconsistency resulting from systematic error from nonphylogenetic signals or model inadequacy, increases when large amounts of data are included in phylogenomic analyses [71]. To minimize this risk, it has been suggested that the number of genes or loci used in phylogenomic analyses should be proportional to the number of taxa being studied [71]. In the context of Armillaria, the need for more resources and techniques for species delimitation has been highlighted in various review articles, including those by Baumgartner et al. [2], Coetzee et al. [3] and Heinzelmann et al. [5]. The availability of genome sequences and the possibility of sequencing more genomes offer more opportunities for bridging this gap. Several genomes of Armillaria spp. with diverse geographical distribution will be required for further phylogenomic and/or phylogeographic studies to fully delimit species and species complexes among Armillaria, detect trends in horizontal gene transfer, decipher other evolutionary history of these fungi, and use multispecies phylogenetic comparisons to infer putative functions for DNA or protein sequences.

Genomic studies have also resulted in deciphering some of the molecular basis of sexual reproduction in Armillaria spp. The mating (MAT) loci of A. solidipes, A. ostoyae, A. cepistipes and A. gallica have been reported [16]. Further insights were gained about the chromosomal locations of the MAT-A and MAT-B loci with the establishment of the only near-chromosome-scale genome of an Armillaria spp., namely A. ostoyae. This research led to the discovery of the location of the MAT loci on pseudochromosomes LG1 and LG6, respectively [15]. Fungal sexual reproduction influences genetic recombination, with a consequent effect on genetic diversity. These mechanisms can play a role in the evolutionary ‘arms race’ between fungal phytopathogens and their hosts, as well as adaptation to hostile environments [72,73]. Hence more information about the MAT loci is relevant for understanding some of the mechanisms underlying the broad host range, host–pathogen interactions and global distribution of Armillaria spp.

Discoveries made about some important characteristics of Armillaria spp. have benefited from multidisciplinary approaches which incorporated genomics with other experimental investigations. For instance, through multi-omics studies including transcriptomics and proteomics, Sipos et al. reported that A. cepistipes, A. ostoyae and A. gallica possess more pectinolytic gene families than lignolytic gene families compared with other white-rot fungi [16]. The authors suggest that these gene families may enable Armillaria spp. to avoid competition with other white-rot fungi, enhancing their ability to colonize and thrive on some hosts. Differential regulation of the genes encoding these biotic factors was also reported in various tissues (mycelia, rhizomorphs, fruiting bodies at different developmental stages and in different tissues of a mature fruiting body) of A. ostoyae, with a potential influence on multicellular development as well as cap and stipe differentiation [16].

Other important discoveries have been made about Armillaria spp. based on multi-omics and biological investigations. Combining genomics with in vitro analyses of recombination rate variation, Heinzelmann et al. discovered 1984 crossover events in the progeny of A. ostoyae, highly variable recombination rates along chromosomes, and regions close to the telomeres serving as gene-poor recombination hotspots [15]. Multiple glycol-degradative enzyme systems which can be useful for recalcitrant carbon compound degradation, and potent antifungal properties of A. mellea, were also discovered based on genomic and proteomic studies [17]. Moreover, comparative genomics coupled with in vitro targeted metabolomics studies resulted in records of conserved putative nonribosomal peptide synthetases-dependent siderophore synthetase gene clusters and in vitro species- and strain-independent biosynthesis of different types and quantities of siderophores by strains of various Armillaria species [74]. Hence, while genomic data can provide some answers, using genome data in combination with in vitro, in situ and in plantae studies, including other bioinformatics studies, can provide a much greater understanding of the species in this genus. This multidisciplinary approach to investigating fungal phytopathogens has also been recommended in the review by Aylward et al. [75].

Suggestions for future genomics & transcriptomics studies on the genus Armillaria

Armillaria genome projects have yielded 11 high-quality publicly available genomes at the contig or scaffold levels on the NCBI database. These genomes have been applied in further research to gain a better understanding of these fungi and of phytopathogens in general. Nonetheless, there is still more to be achieved in terms of genomics and multi-omics research on Armillaria spp. In this section, prospects in Armillaria genomics are suggested.

So far, the genome of A. ostoyae strain SBI C18/9 is the only Armillaria genome that has been assembled to pseudochromosome (near-chromosome) level. The pseudochromosome was constructed based on the order of scaffolds within genetic map linkage groups of haploid progenies of A. ostoyae strain C15 [15]. Efforts should be made toward obtaining more chromosome-scale Armillaria genomes in order to derive the benefits of such data, as has been achieved for vertebrate genomes using pipelines such as the Vertebrate Genomes Project assembly pipeline, which was designed for generating high-quality, near-error-free, gap-free, chromosome-level, haplotype-phased, annotated reference genome assemblies for every vertebrate species [76]. This pipeline is hosted on the Galaxy platform and uses PacBio or Nanopore reads and additional techniques such as Hi-C sequencing [60,76]. Hi-C sequencing is a supporting technology used to improve the contiguity of genome assemblies [27,77]. It is expected that the Vertebrate Genomes Project pipeline will be suitable for assembling diploid, highly heterozygous Armillaria genomes to chromosomal level and this should be explored in future. It is imperative that this pipeline be validated using raw long- and short-read data of Armillaria sequences as well as additional sequences obtained using other sequencing techniques. In the absence of Hi-C and other supporting data, the workflows in Figures 1 & 2 can be used as guidelines for embarking on more Armillaria genome projects to achieve optimum genome assemblies and annotation given available resources. Resources for accessing the tools indicated in Figures 1 & 2 as well as other resources have been provided in Supplementary Table 1.

Other discoveries can be achieved with more studies based on Armillaria genomes. These discoveries can be facilitated by combining genomics with other experimental studies. The benefits of availability of genomes of numerous strains of fungal phytopathogens coupled with other wet-bench experimental studies have been discussed [75,78]. In the context of Armillaria, embarking on assembling more high-quality genomes of several strains belonging to various species and with a worldwide geographical sampling promises to be highly resourceful, especially if a multidisciplinary approach is adopted. Such studies will help bridge the wide knowledge gap about the mechanisms used by these fungi in their evolutionary success. Additionally, the considerable variation in aggressiveness, pathogenicity and virulence both intra- and inter-species among pathogenic members of Armillaria is still not clearly understood [25,26]. The mechanisms underlying these characteristics can also be deciphered with the multidisciplinary approach.

Some other possible research areas that can be facilitated by the multidisciplinary approach include:

1.	Putatively establishing genes encoding enzymes involved in secondary metabolite biosynthesis and deciphering the expression patterns of genes encoding secondary metabolites in vegetative mycelia, rhizomorphs and fruiting bodies at different developmental stages of different Armillaria spp. under different growth conditions. This will help further understand the role that different developmental stages and different tissues play in Armillaria pathogenicity.
2.	Further research aimed at establishing and understanding the mechanisms of the antibacterial, antifungal and cytotoxic properties of secondary metabolites produced by Armillaria spp.
3.	Further investigation of the molecular basis of the mating system of Armillaria spp. or strains; for instance, to unravel the mechanisms of diploidization of the haploid mycelia and the signals that trigger these mechanisms.
4.	Developing and/or optimizing tools for functional validation of genes in Armillaria for the study of gene function and annotation in Armillaria.
5.	Conducting more research on the secretion, presence, structure and activity of ligninolytic enzymes, other plant cell wall degrading enzymes (PCWDEs) and other carbohydrate-active enzymes of Armillaria spp. in silico, in vitro and in situ.
6.	Comparing the quantity and diversity of pathogenicity-related gene products such as PCWDE in Armillaria spp. with various lifestyles in silico, in vitro and in plantae to establish the effect of enzyme machineries possessed by different Armillaria spp. on their pathogenicity and how they circumvent or outcompete other sympatrically existing Armillaria spp. as well as microbes normally used as biocontrol agents of Armillaria root-rot disease. This information can be used for developing enhanced biocontrol methods of Armillaria root-rot.

Conclusion

Armillaria genome projects have achieved several milestones toward understanding the biology and evolution of these fungi. Advantage should be taken of the fast pace of technological advancement in genome sequencing, assembly and annotation, and the reducing cost of these technologies, to increase genome projects in Armillaria and other Basidiomycota. These should be coupled with other wet-bench experimental studies in a timely manner. Results from these studies can afford the scientific, agroindustrial and pharmaceutical sectors the benefits of the knowledge and technologies that can be gained from these projects.

Future perspective

Given the rapid advances in techniques and technologies relating to genome sequencing, assembly and annotation, as well as the continually reducing costs, we expect an exponential increase in the number of genome sequences in the next year or two. The information provided in this review will help facilitate this. The processes described toward producing high-quality Armillaria genome sequences can be extrapolated to genomic studies of other diploid and dikaryotic fungi and other organisms. It is expected that the availability of high-quality genome sequences will enable a better understanding of the biology and biochemistry of organisms when multidisciplinary approaches are followed. This has the potential to result in the discovery of novel biotechnological products with applications in agriculture, bioremediation, forestry and medicine. Consequently, this would facilitate the achievement of several of the 17 Sustainable Development Goals to sustain our world [79].

Executive summary

Background

• The current status of genomic studies on Armillaria species, with a focus on the diversity of species and strains for which genomes have been assembled and annotated, is considered in this comprehensive overview of the research carried out on this fungal genus.

Status of genome sequencing, assembly & annotation in Armillaria

• Various steps, techniques and tools have been used for Armillaria genome extraction, sequencing, assembly, evaluation and annotation.

• Infographic guidelines for future Armillaria genome projects with some relevant resources are provided.

• The guidelines and resources provided can be applied to other fungal genera and beyond mycology.

Knowledge gained from Armillaria genomics

• Armillaria genome projects have provided draft and complete genome assemblies which have been used to gain further insights into the biology and chemistry of these fungi.

• The range of predicted gene contents of the Armillaria genomes are within the range of between 5,000 and 29,000 genes predicted for other Agaricales genomes.

Suggestions for future genomics studies on the genus Armillaria

• Potential areas for future Armillaria genome projects, as well as areas for potential comparative genomics and multi-omics studies, are outlined.

Conclusion

• Although this review centers around genomic studies of Armillaria spp., the methods, guidelines and information gathered from these studies provide valuable information that can be utilized for genomic and transcriptomic research on other fungi and beyond mycology.

Author contributions

MPA Coetzee and BD Wingfield supervised the project; DL Narh Mensah collected the information, designed the tables and figures and wrote the manuscript. All authors read and agreed to publish the final version of the manuscript.

Supplementary data

To view the supplementary data that accompany this paper please visit the journal website at: www.tandfonline.com/doi/suppl/10.2144/btn-2023-0023

Financial disclosure

Funding was received from the University of Pretoria, the South African Department of Science and Innovation (DSI)-National Research Foundation (NRF) Centre of Excellence in Plant Health Biotechnology (CPHB, previously the CTHB), the DSI–NRF South African Research Chairs Initiative (SARChI) in Fungal Genomics (grant no. 98353) and the Tree Protection Cooperative Program (TPCP). The grant holders acknowledge that opinions, findings and conclusions or recommendations expressed in this publication are those of the authors, and that the funding bodies accept no liability whatsoever in this regard. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Competing interests disclosure

The authors have no competing interests or relevant affiliations with any organization or entity with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

Writing disclosure

No writing assistance was utilized in the production of this manuscript.

Additional information

Funding

Funding was received from the University of Pretoria, the South African Department of Science and Innovation (DSI)-National Research Foundation (NRF) Centre of Excellence in Plant Health Biotechnology (CPHB, previously the CTHB), the DSI–NRF South African Research Chairs Initiative (SARChI) in Fungal Genomics (grant no. 98353) and the Tree Protection Cooperative Program (TPCP). The grant holders acknowledge that opinions, findings and conclusions or recommendations expressed in this publication are those of the authors, and that the funding bodies accept no liability whatsoever in this regard. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.