Micro-and minisatellites are among the most powerful genetic markers known to date. They have been used as tools for a large number of applications ranging from gene mapping to phylogenetic studies and isolate typing. However, identifying micro-and minisatellite markers on large sequence data sets is often a laborious process.
FONZIE was designed to successively 1) perform a search for markers via the external software Tandem Repeat Finder, 2) exclude user-defined specific genomic regions, 3) screen for the size and the percent matches of each relevant marker found by Tandem Repeat Finder, 4) evaluate marker specificity (i.e., occurrence of the marker as a single copy in the genome) using BLAST2.0, 5) design minisatellite primer pairs via the external software Primer3, and 6) check the specificity of each final PCR product by BLAST. A final file returns to users all the results required to amplify markers. A biological validation of the approach was performed using the whole genome sequence of the phytopathogenic fungus Leptosphaeria maculans, showing that more than 90% of the minisatellite primer pairs generated by the pipeline amplified a PCR product, 44.8% of which showed agarose-gel resolvable polymorphism between isolates. Segregation analyses confirmed that the polymorphic minisatellites corresponded to single-locus markers.
FONZIE is a stand-alone and user-friendly application developed to minimize tedious manual operations, reduce errors, and speed up the search for efficient minisatellite and microsatellite markers departing from whole-genome sequence data. This pipeline facilitates the integration of data and provides a set of specific primer sequences for PCR amplification of single-locus markers. FONZIE is freely downloadable at: http://www.versailles-grignon.inra.fr/bioger/equipes/leptosphaeria_maculans/outils_d_analyses/fonzie webcite
Satellite sequences are abundantly interspersed in the genome of almost all eukaryotic species studied  and have been analyzed extensively in animals [2-4], plants , and more recently in fungi .
Two major classes of satellites are usually defined: minisatellites (MS) and microsatellites (μS). μS are repetitive sequences of mostly 2 to 4 nucleotides with a widespread occurrence in multicellular organisms, whereas minisatellites are usually defined as the repetition in tandem of a short, 6 to 100 bp motif, spanning 100 bp to several kilobases .
μS analysis requires sophisticated separation and visualization apparatus, due to the small size of the individual repeat units. In contrast, and because of the larger size of their individual core motif, MS can be separated and visualized by conventional agarose gel electrophoresis . Using a panel of single-locus minisatellite markers, isolate-specific DNA fingerprint can be produced in a PCR-based assay [8,9]. MS also provided highly polymorphic markers for linkage studies  making them informative genetic markers.
Genome mapping is still the primary tool for genome knowledge in a series of model organisms. Due to their small genome size and laboratory tractability, fungal models such as the ascomycete yeast Saccharomyces cerevisiae, and the filamentous ascomycete Neurospora crassa, have been pioneers in the process of building genetic maps . In this respect, the Dothideomycete Leptosphaeria maculans causing stem canker of oilseed rape (Brassica napus), is amenable to genetics in the lab and is thus an adequate and complete model for genome-wide-based functional studies of pathogenicity, and for identifying signalling and regulation processes responsible for shifts in lifestyle . Although preliminary genetic maps have been developed for L. maculans [13-15], it is difficult to integrate these maps as the markers (RAPD and AFLP) used to generate them are not readily transferable . Based on the analysis of a set of around 800 BAC-end sequence data, MS were found to be powerful for genetic mapping or population genetic studies in L. maculans . The availability of the L. maculans genome now enables the generation of a large quantity of such markers, which could allow the saturation of the genetic map and improve population genetic studies and pathogenicity gene discovery .
Despite the availability of whole genome sequences, in silico MS identification and primer design is still a laborious process, requiring the use of different softwares and numerous steps of marker validation, which take time and often generate errors. Most of the currently available software or pipeline solutions are focused on the identification of μS [17-19]. Moreover, they are usually platform dependent , or may need several tedious pre-processing steps [19,21]. In order to avoid these constrains and to reduce time, FONZIE has been developed to automate and facilitate the design of PCR primers from large sets of sequences, that will amplify single locus MS and their flanking sequences, with the possibility of excluding some specific regions of the genome from the MS design process. This pipeline integrates various external tools such as BLAST , Tandem Repeat Finder (TRF) , and Primer3 . FONZIE is able to successively perform TRF on a Fasta or Multifasta formatted sequence file, search for markers specifically located in user-defined sequence subset, as exemplified here with GC-equilibrated isochores of L. maculans, eliminate markers which do not satisfy user-defined criteria, eliminate markers which are not single-copy, and finally design minisatellite primer pairs and check the specificity of the PCR product. The primary targeted public of FONZIE is biologists unfamiliar with complex bioinformatics solutions. Therefore it does not require a computer science background and can run on personal computers, yielding results within an hour or less for most queries.
A comprehensive "readme" page includes details on how to configure and run the pipeline, and how to search and display FONZIE results.
Before launching FONZIE, users must have installed the Python programming language on the computer. In a second step, users must create a database of sequences (ideally, the whole-genome sequence of the investigated organism) in order to perform BLAST during the execution of the pipeline. In our project, the database used was the pilot genome of L. maculans (45.12 Mb, 76 Super-Contigs, Table 1) as a Multifasta file.
Table 1. FONZIE results when performed on different fungal or oomycetes whole genomes.
FONZIE consists of three major components: a set of pipelined programs, a BLAST database and a graphical user interface (Figure 1). The FONZIE pipeline consists of Python modules allowing the execution of different external software programs.
Figure 1. The graphical user interface of FONZIE. The graphical user interface is composed of five major sections. a. TRF parameters: All the modifiable parameters of TRF are located in this section. Users can modify each of them separately. b. Screening parameters. Users can modify the core motif size and the percent match between core motifs c. BLAST cut-off parameters: Users can modify the two E-value cut-off used during the screening for marker specificity and the virtual PCR steps d. Flanking sequences size and Primer3 parameters. e. The file field: In this section, users can open the Fasta- or Multifasta formatted sequence file, the BLAST database file, and the optional gff format file which contains the excluded regions.
The BLAST database was created with FORMATDB, a BLAST  database-related tool. The graphical user interface was implemented with a Python graphic standard library named Tkinter. FONZIE was developed under both Linux Ubuntu 7.10 and Windows XP operating systems.
The automated process consists of six steps (Figure 2):
Figure 2. The FONZIE workflow: The FONZIE workflow consists of 6 steps. a Execution of TRF on a Fasta sequence or a Multifasta file, b Optional: exclusion of some user-defined specific regions, c Screening by core motif size and percent match parameters, d Screening for marker specificity with BLAST (1e-10 by default) against a database, e Primer design step using PRIMER3, f Virtual PCR by making a BLAST (1e-40 by default) against the database to check the specificity of the PCR product (single copy locus).
Tandem Repeat Finder
FONZIE accepts nucleotidic sequence files in Fasta or Multifasta format and uses the Tandem Repeat Finder software  in order to find tandemly repeated elements (minisatellites and microsatellites) (Figure 2 step a). TRF parameters can be modified via the graphical user interface (Figure 1).
Exclusion of markers in specific regions
Isochores have been described for many eukaryotic organisms such as plants and mammals [25,26]. Isochores are very long stretches of DNA that are homogeneous in base composition and are compositionally correlated with the coding sequences that they embed. In the L. maculans genome this structure is particularly clear, with alternation of GC-equilibrated coding regions with no or few transposable elements (TE) and AT-rich regions mainly composed of degenerated and truncated TE and encompassing very few ORFs [13,27]. Due to these particularities, most micro-or minisatellites identified in L. maculans AT isochores are not single locus markers and have to be excluded from the search for single locus markers. This can be easily extended to all repeat-rich genomes such as that of the Oomycete Phytophthora infestans containing 74% of repeats .
In order to exclude such specific genomic regions from the search for markers, FONZIE compares the location of the tandem repeats with that of the AT-rich isochores (or whichever other user-defined sequences). Tandem repeats fully or partially overlapping the AT-rich isochores (or other user-defined sequences) are thus automatically excluded from the analysis (Figure 2 step b). The AT-rich isochores or other user-defined sequences to be excluded must be specified in a GFF format file. If no isochore regions, or any other regions to be excluded from the search for MS, are specified, FONZIE analyses all the tandem repeats found by TRF.
Length and percent match screening
Markers are then screened for the size of the core motif (more than 6 bp by default), and the percent matches between core motifs (i.e., the percentage of matched bases between tandem repeats of the core motif within the MS, more than 90% by default). A minimum of 90% identity between core motifs was selected here because previous experiments showed that it was a characteristic of minisatellites showing actual sequence length polymorphism in L. maculans [; unpublished data]. The output of this step is a Fasta formatted sequence file for each selected marker (Figure 2 step c).
Screening for specificity of the markers
This step verifies the specificity of the selected MS (Figure 2 step d). Each MS is analysed for sequence similarity using BLAST  against a BLAST database, created by users, with a default E-value cut-off set at 1e-10. This value can be modified via the graphical user interface. Typically, the specificity is evaluated against the whole genome sequence if available.
Using the output of the BLAST, FONZIE classifies MS into 4 categories, defined by both the number of sequences in the BLAST database matching the query sequence (i.e. with BLAST E-value < 1e-10) and the E-value obtained for the query sequence compared to that obtained for other matching sequences, as follows:
1) MS for which the BLAST results exhibit a unique match in the database, corresponding to the query sequence, hereafter referred to as "UNIQUE COPY".
2) MS for which the BLAST results exhibit several matches on several sequences, with a best hit on the query sequence, hereafter referred to as "MULTIPLE COPIES".
3) MS for which the BLAST results exhibit several matches on several sequences, with a best hit on other sequence than the query sequence, hereafter referred to as "MULTIPLE COPIES (BEST HIT ON SEQUENCE X)". X represents the sequence id. with the best alignment score. The repetition of the motifs among the whole genome can explain such results.
4) (Rare) MS for which the BLAST results exhibit no match on any sequences with the default cut-off of E-value at 1e-10, hereafter referred to as "NO BLAST RESULTS" (0.2% in the case of L. maculans, Table 1). The high default threshold parameter of the E-value (1e-10), the low complexity of the motifs, or the small size of the MS sequence can explain such results.
"MULTIPLE COPIES (BEST HIT ON SEQUENCE X)" and "NO BLAST RESULTS" are automatically excluded for the next steps of the analysis.
In order to design primer pairs flanking the MS sequences, FONZIE extracts 100 bps by default on each side of the query sequence whenever possible (or whichever other user-defined length, modified via the graphical user interface, Figure 1). Otherwise, FONZIE extracts the maximum number of nucleotides available upstream and downstream of the MS sequence.
FONZIE then uses Primer3  which analyses the flanking DNA sequences in order to define suitable forward and reverse PCR primers, designed with the standard set of constraints of Primer3 (Figure 2 step e): optimal primer size at 20 bases pairs, optimal Tm at 60°C, and optimal primer GC content at 50%. These parameters can also be modified by users via the graphical user interface (Figure 1). FONZIE then calculates the final product size range as the size of the MS (= the size of the core motif × the number of repeats as defined by TRF) + 2×, or 10×, the minimal primer size (chosen by users), for the minimal and the maximal size, respectively. The first primer pair returned by Primer3 is displayed by FONZIE. If no primer pairs are found with Primer3, the result is displayed as "NO DESIGNED PRIMER" in the final text file.
The virtual PCR step validates the specificity of amplification products identified in step e (Figure 2 step f). Each amplification product is analysed for sequence similarity using BLAST against all the sequences of the BLAST database. The E-value cut-off is fixed at 1e-40 by default and may be modified by the users. The same type of analysis is implemented in the SAT software , but SAT only verifies the specificity of the PCR primers, whereas FONZIE analyses the specificity of the whole amplification product.
Using the outputs of the virtual PCR step, FONZIE classifies sequences in 4 categories according to the specificity of amplification products. These categories are exactly the same as for the markers' specificity screen.
Running the pipeline
Users have to provide an input sequence (Fasta or Multifasta format), and a GFF format file if they want to exclude specific regions from the analysis.
Setting parameters and pipeline execution
TRF default options are as follows: alignment parameters of 2,7,7, which correspond to the weight attributed to each match, mismatch and indel, respectively, between core motifs of the MS; minimum alignment score of 50; maximum period size of 500, pM (i.e., the average percent identity between the copies of a pattern) of 80; pI (i.e., the average percent of insertions and deletions between the copies of a pattern) of 10 (Figure 1) . Users can modify through the graphical interface all TRF parameters, but also the screening parameters (i.e., the minimum identity between core motif fixed by default at 90%, and the size of the motif, >6 bps by default) as well as the two BLAST cut-off parameters (E-values cut-off of 1e-10 and 1e-40 by default) (Figure 1).
Visualization of results
At the end of the process, users have access to all the files and directories created at each step of the pipeline. A final tab-delimited text file summarizes, for all numbered satellites found by FONZIE, the name of each satellite automatically attributed by the pipeline, the core motif of each satellite, the number of repetitions of the core motif, the BLAST results of specificity of the satellite and of the amplification product (i.e. if the marker and the amplification product are found as a unique copy or not), the sequences of the forward and reverse primers and the start and the end location of the amplification product on the input sequence (Table 2).
Table 2. Example of the FONZIE final result table, run on Supercontig 16 of the Leptosphaeria maculans genome.
Sequence files and directory
FONZIE allows traceability of each step of the pipeline by generating a directory and associated files. FONZIE displays a DAT file created by TRF which contains the information of all tandem repeats detected on the input sequence. A text file shows the different consensus sequences of each marker, its position on the input sequence as well as period size, copy number, consensus size, percent matches, percent indels, the score and the status of each marker.
A directory named "MARKERS_SEQUENCES" is created at the end of the first screening step which contains all the markers found by FONZIE, in Fasta format. Another directory named "AMPLIF_PRODUCTS_SEQUENCES", contains the Fasta files for the sequence of each amplification product. A last directory named "MARKERS_RECAP" contains, for each MS, a text file summarizing the marker sequence, the marker sequence with flanking regions, the amplification product sequence, the amplification product status, the positions of the amplification product on the query sequence, and the output of Primer3 indicating the characteristics of the primers and their location in relation to the MS (Figure 3).
Figure 3. Example of one file from "MARKERS_RECAP" directory generated by FONZIE. From top to bottom: the marker sequence, the marker sequence with flanking regions, the amplification product sequence, and the primer3 output
Results and discussion
FONZIE was firstly assessed on the whole L. maculans genome. The 45 Mbs were analysed in 23 minutes (Table 1). FONZIE found 2606 minisatellites, with 2405 amplification products occurring as unique copies, 146 in multiple copies. Six amplification products had no BLAST results, and 49 markers were identified for which no primers could be designed. 92% of the markers identified by FONZIE, following exclusion of repeat-rich AT-rich isochores thus corresponded to putatively unique sequences in the genome and could be further used for analysis of polymorphism in isolates and genetic mapping and/or population genetics studies.
In a second step, FONZIE was used to mine 6 other fungal and Oomycete genomes. These genomes, sized from 34 Mb to 228 Mb (Table 1), range from compact, repeat poor genomes (Aspergillus niger, Stagonospora nodorum, Pyrenophora tritici repentis and Botrytis cinerea) to repeat rich genomes of the basidiomycete Laccaria bicolor  and the Oomycete P. infestans . MS were found using the same parameters as for L. maculans, but without defining user-specified genomic regions to be excluded from the analysis. In all cases, numerous MS could be identified and extracted in a very limited time interval, i.e. between 2 minutes to cover the A. niger whole genome to 3 h21 min to scan the P. infestans genome sized 228 Mb over 4921 contigs (Table 1). This analysis suggests some genomes are extremely rich in MS (e.g., the B. cinerea genome; Table 1). It mainly indicates that an average of 70-83% single-copy markers are identified whenever the genomes investigated are poor in repeats or when repeat-rich regions are excluded from the analysis, as done here for L. maculans (Table 1). In contrast, only 23-30% of markers were found to be single-copy markers in the genomes of the repeat-rich fungi and Oomycete L. bicolor and P. infestans if analysing the whole genomes with FONZIE, without user-defined exclusion of repeat-rich sequences.
For validation of the proposed approach, primer pairs for 517 putative single-copy MS markers (i.e. with core motif ≥6), scattered along 17 L. maculans supercontigs (with AT-rich isochores as excluded regions), were PCR-assayed on a set of 9 L. maculans natural isolates. More than 90% of the primer pairs (475) amplified a PCR product, with size compliant with the expected MS size. This yield is very similar to that of the currently available tools such as STAMP, yielding 83.33 to 91.67% of successful amplifications, depending on the species . Among these PCR-amplified sequences, 44.8% (213 primer pairs) showed resolvable size polymorphism between at least 2 of the 9 tested L. maculans isolates. Finally, 24.8% of the 517 MS (i.e. more than 55% of the polymorphic MS markers) were found to be polymorphic between isolates 'a.2' and 'H5', the two parental isolates of the L. maculans genetic map . Progeny analysis confirmed that all these markers segregated as single loci, i.e. with a 50:50 ratio, as expected for this haploid organism.
In previous experiments, MS markers were searched for in L. maculans by using the TRF application alone , with a yield of only 13% of polymorphic minisatellites out of 30 primer pairs tested, even though a wider range of isolates (21 isolates instead of 9 in the present study) was used for the polymorphism screen. Only 10% of the primer pairs from this preliminary study were found to be polymorphic between isolates 'a.2' and 'H5', compared to 24.8% in the present work. These results not only validate the efficiency of the pipeline but also confirm the potential of MS as genetic markers in L. maculans, since it can be extrapolated both from Table 1 and these results that more than 530 MS could be mapped to the reference L. maculans map.
FONZIE is a user-friendly application, optimised for MS marker identification and specific PCR primer design. It brings together external software like Tandem Repeat Finder, Primer3 and BLAST, to perform them in one run for PCR primer pair design. FONZIE is time-saving, reduces errors that might be introduced by analysing sequence by hand and is easy to use. Users only need to upload their sequences, a reference database, an optional file containing excluded genomic regions, and select appropriate parameters. Moreover, FONZIE can cover whole genomes in one run. This tool is generic enough to be used on the DNA of any organism and is particularly well adapted to whole genome sequence from which specific regions (AT-isochores for examples) need to be excluded.
The main innovation of the pipeline is to integrate two successive steps of BLAST to check the specificity and uniqueness of the MS loci identified by TRF. The usefulness of the pipeline has been demonstrated here for the fungus L. maculans, since 100% of the polymorphic markers identified were single locus, and the yield of identification of polymorphic markers has been more than doubled compared to manual search for MS without BLAST steps.
Availability and requirements
• Project name: FONZIE
• Operating system: UNIX/LINUX, Windows XP
• Programming language: Python
• Licence: GNU GPL, free for academic and non-academic users
• Any restrictions to use by non-academic: none
The authors declare that they have no competing interests.
PB and JG designed the application, JG programmed the application and supervised the implementation of the computational part of this work, PB and JG tested and evaluated the application, TR contributed to validation on fungal genomes and edited the manuscript, PB and MHB performed the biological validation of the markers on L. maculans, PB wrote the manuscript, MHB initiated and coordinated the work and helped to draft the manuscript. All authors read and approved the final manuscript.
This work and PB were funded by the French agency ANR ("Agence Nationale de la Recherche") contract ANR-07-GPLA-015 ("AVirLep") under the framework of the Génoplante 2010 programme. JG was funded by the ANR project "Fungeffector" (ANR-06-BLAN-0399) and by the INRA SPE (Santé des Plantes et Environnement) department. The authors thank Siân Deller (INRA BIOGER) for providing useful comments on the manuscript.
Dharma Prasad M, Muthulakshmi M, Madhu M, Archak Sunil, Mita K, Nagaraju J: Survey and analysis of microsatellites in the silkworm, Bombyx mori: frequency, distribution, mutations, marker potential and their conservation in heterologous species.
Haas BJ, Kamoun S, Zody MC, Jiang RHY, Handsaker RE, Cano LM, Grabherr M, Kodira CD, Raffaele S, Torto-Alalibo T, Bozkurt TO, Ah-Fong AMV, Alvarado L, Anderson VL, Armstrong MR, Avrova A, Baxter L, Beynon J, Boevink PC, Bollmann0 SR, Bos JIB, Bulone V, Cai G, Cakir S, Carrington JC, Chawner M, Conti L, Costanzo S, Ewan R, Fahlgren N, Fischbach MA, Fugelstad J, Gilroy EM, Gnerre S, Green PJ, Grenville-Briggs LJ, Griffith J, Grünwald NJ, Horn K, Horner NR, Hu CH, Huitema E, Jeong DH, Jones AME, Jones JDG, Jones RW, Karlsson EK, Kunjeti SG, Lamour K, Liu Z, Ma LJ, MacLean D, Chibucos MC, McDonald H, McWalters J, Meijer HJG, Morgan W, Morris PF, Munro CA, O'Neill K, Ospina-Giraldo M, Pinzón A, Pritchard L, Ramsahoye B, Ren Q, Restrepo S, Roy S, Sadanandom A, Savidor A, Schornack S, Schwartz DC, Schumann UD, Schwessinger B, Seyer L, Sharpe T, Silvar C, Song J, Studholme DJ, Sykes S, Thines M, van de Vondervoort PJI, Phuntumart V, Wawra S, Weide R, Win J, Young C, Zhou S, Fry W, Meyers BC, Van West P, Ristaino J, Govers F, Birch PRJ, Whisson SC, Judelson HS, Nusbaum C: Genome sequence and analysis of the Irish potato famine pathogen Phytophthora infestans.
Martin F, Aerts A, Ahrén D, Brun A, Danchin EG, Duchaussoy F, Gibon J, Kohler A, Lindquist E, Pereda V, Salamov A, Shapiro HJ, Wuyts J, Blaudez D, Buée M, Brokstein P, Canbäck B, Cohen D, Courty PE, Coutinho PM, Delaruelle C, Detter JC, Deveau A, DiFazio S, Duplessis S, Fraissinet-Tachet L, Lucic E, Frey-Klett P, Fourrey C, Feussner I, Gay G, Grimwood J, Hoegger PJ, Jain P, Kilaru S, Labbé J, Lin YC, Legué V, Le Tacon F, Marmeisse R, Melayah D, Montanini B, Muratet M, Nehls U, Niculita-Hirzel H, Oudot-Le Secq MP, Peter M, Quesneville H, Rajashekar B, Reich M, Rouhier N, Schmutz J, Yin T, Chalot M, Henrissat B, Kües U, Lucas S, Van de Peer Y, Podila GK, Polle A, Pukkila PJ, Richardson PM, Rouzé P, Sanders IR, Stajich JE, Tunlid A, Tuskan G, Grigoriev IV: The genome of Laccaria bicolor provides insights into mycorrhizal symbiosis.
Kraemer L, Beszteri B, Gabler-Schwarz S, Held C, Leese F, Mayer C, Pohlmann K, Frickenhaus S: STAMP: Extensions to the STADEN sequence analysis package for high throughput interactive microsatellite marker design.