A framework physical map for peach, a model Rosaceae species

Tree Genetics & Genomes DOI 10.1007/s11295-008-0147-z

ORIGINAL PAPER

A framework physical map for peach, a model Rosaceae species T. N. Zhebentyayeva & G. Swire-Clark & L. L. Georgi & L. Garay & S. Jung & S. Forrest & A. V. Blenda & B. Blackmon & J. Mook & R. Horn & W. Howad & P. Arús & D. Main & J. P. Tomkins & B. Sosinski & W. V. Baird & G. L. Reighard & A. G. Abbott

Received: 23 July 2007 / Revised: 8 January 2008 / Accepted: 21 February 2008 # Springer-Verlag 2008

Abstract A genome-wide framework physical map of peach was constructed using high-information content fingerprinting (HICF) and FPC software. The resulting HICF assembly contained 2,138 contigs composed of 15,655 clones (4.3× peach genome equivalents) from two complementary bacterial artificial chromosome libraries. The total physical length of all contigs is estimated at 303 Mb or 104.5% of the peach genome. The framework physical map is anchored on the Prunus genetic reference map and integrated with the peach transcriptome map. The physical length of anchored contigs is estimated at 45.0 Mb or 15.5% of the genome. Altogether, 2,636 markers, i.e., genetic markers, peach unigene expressed sequence tags, and gene-specific and overgo probes, were incorporated into the physical framework and supported the accuracy of contig assembly. Communicated by R. Velasco Electronic supplementary material The online version of this article (doi:10.1007/s11295-008-0147-z) contains supplementary material which is available to authorized users. T. N. Zhebentyayeva (*) : L. L. Georgi : L. Garay : S. Forrest : A. V. Blenda : J. Mook : R. Horn : A. G. Abbott Department of Genetics and Biochemistry, Clemson University, Clemson, SC 29634-0318, USA e-mail: [email protected] G. Swire-Clark : W. V. Baird : G. L. Reighard Department of Horticulture, Clemson University, Clemson, SC 29634-0319, USA B. Blackmon : J. P. Tomkins Clemson University Genomics Institute, Clemson University, Clemson, SC 29634, USA S. Jung : D. Main Department of Horticulture and Landscape Architecture, Washington State University, Pullman, WA 99164-6414, USA

Keywords Prunus persica . HICF fingerprinting . BAC-based physical/genetic map

Introduction The Rosaceae, a large and economically important angiosperm family, comprises more than 3,000 species in approximately 110 genera distributed worldwide (Takhtajan 1997). Members of the Rosaceae exhibit considerable diversity in terms of growth habit, leaf morphology, and fruit type. Based on the fruit type, which is closely correlated with chromosome number, the Rosaceae is traditionally divided into four well-defined subfamilies Maloideae (x=17), Prunoideae (x=8), Rosoideae (x=7, 8, W. Howad : P. Arús Centre de Recerca en Agrigenòmica CSIC–IRTA–UAB, Carretera de Cabrils Km 2, 08348, Cabrils, Barcelona, Spain B. Sosinski Department of Horticultural Sciences, North Carolina State University, Raleigh, NC 2769-8619, USA J. Mook Agricultural Research Programs, Virginia State University, P.O. Box 9061, Petersburg, VA 23806, USA R. Horn Abt Pflanzengenetik, Institut für Biowissenschaften, Universität Rostock, Albert-Einstein-Str 3, Rostock 18051, Germany

Tree Genetics & Genomes

or 9), and Spiraeoideae (x=9) (Potter 2003). Due to a twofold difference in chromosome number and other molecular evidence, the Maloideae is now being treated as an ancient polyploid resulting from autopolyploidy within the Spiraeoideae subfamily rather than from allopolyploidy involving Spiraeoideae and Prunoideae lineages (Gladkova 1972; Evans and Campbell 2002; Potter 2003). Because of its ecological, morphological, and karyotypic diversity, the Rosaceae provides numerous opportunities for evolutionary studies. These include such questions as basic chromosome number variation and character variation associated with these changes, fruit type and plant habit evolution accompanied by metabolic pathway specialization, and microbe– plant co-evolution leading to actinorhizal species. To facilitate investigation of these fundamental and applied questions, genetic and genomic resources for economically important rosaceous crops have been developed and organized into an integrated web-based Genome Database for Rosaceae or GDR (Jung et al. 2004; http://www.bioinfo.wsu. edu/gdr). The GDR provides genetic linkage maps, bacterial artificial chromosome (BAC) library resources, and expressed sequence tag (EST) databases for the Maloideae (apple and pear), Rosoideae (strawberry, raspberry, and rose) and Prunoideae (peach, apricot, plum, cherry, and almond). Also, an initial physical/genetic framework map for peach (Zhebentyayeva et al. 2006) and recently published physical map for apple (Han et al. 2007) are available to the Rosaceae scientific community. Peach [Prunus persica (L.) Batsch], a member of the Prunoideae subfamily, is both one of the most widely grown and the best genetically characterized species in the Rosaceae. As a result of its small genome size, the developing genomic resources, and the colinearity of the peach genome within Prunus, diploid peach has become a model genome for fruit crops providing a platform for comparative studies within the entire Rosaceae (Abbott et al. 2002). To date, 15 molecular genetic maps have been constructed for peach and other Prunus species (http:// www.bioinfo.wsu.edu/gdr). Map comparisons using transferable genetic markers demonstrated that the studied diploid Prunus species (almond, peach, apricot, cherry, and plum) share nearly identical genome organization (Arús et al. 2006; Abbott et al. 2006). A stepwise saturated linkage map for the almond ‘Texas’×peach ‘Earlygold’ F2 population (T×E map) is recognized as a general Prunus map (Joobeur et al. 1998; Aranzana et al. 2003; Dirlewanger et al. 2004). The T×E map has 562 codominant markers including 11 isozymes, 185 simple sequence repeats (SSRs), 361 restriction fragment length polymorphisms (RFLPs), and five sequence-tagged sites (STSs). It consists of eight linkage groups, the same as the basic haploid chromosome number of the genus, and covers a genetic distance of 519 cM with an average density of 0.92 cM

per marker. Subsequently, Howad et al. (2005) placed 264 additional SSRs on the map using a “bin mapping” approach. On the T×E map, Dirlewanger et al. (2004) established the positions of 28 major genes affecting agronomic characters found in different Prunus species. The fine genetic mapping of morphological traits affecting flower and fruit in peach (Dirlewanger et al. 2006) and quantitative trait locus (QTL) mapping of agronomic traits in almond (Sánchez-Pérez et al. 2007) were initiated recently. Development of peach as a model genome for Rosaceae combines structural and functional genomics efforts that focus on developing a candidate gene database and on constructing an integrated physical/genetic map. Three large-insert genomic libraries and four organ-specific cDNA libraries were made to support the Prunus genomics initiative (http://www.bioinfo.wsu.edu/gdr). Also, a putative unigene set was defined to facilitate hybridization-based assignment of ESTs to physical BACs. Using core RFLP markers from the general Prunus genetic map, BAC clones were anchored on the genetic map, providing a framework for the construction of an initial transcript map (Horn et al. 2005). Lalli et al. (2005) enriched this framework with resistance gene analogs from apricot and peach. Integrated physical/genetic maps are of critical importance for high-throughput EST mapping, QTL fine-mapping, and effective positional cloning of genes (Zhang and Wing 1997; Zhang and Wu 2001; Green 2001). To construct the physical map for peach, we employed essentially the strategies utilized to develop the physical maps for Arabidopsis thaliana and Drosophila melanogaster (Marra et al. 1999; Hoskins et al. 2000). The approach combines hybridization of the genetically mapped markers with BAC DNA fingerprinting and, in our case, hybridization of EST sequences as well. Manual sequencing gel-based fingerprinting is a reliable and cost-effective technique for BAC fingerprinting and, under certain circumstances, performs better than other traditional fingerprinting methods (Xu et al. 2004). An initial acrylamide gel-based physical framework for peach was established and released recently (Zhebentyayeva et al. 2006). This framework was based on random fingerprinting of 3× peach genome equivalents, covered at least 50% of the genome, and included hybridization data for 673 of 3,384 ESTs of the peach unigene set (PP_LEa, i.e., Prunus persica ‘Loring’ fruit ESTs). On this framework, genetically anchored BAC contigs provided landmarks for a Prunus–Arabidopsis microsynteny study (Jung et al. 2006) and for further development of the Prunus transcript map by Horn et al. (2005). Since the first release, the peach physical framework has undergone further enhancements. We integrated global hybridization data that include additional ESTs, amplified fragment length polymorphisms (AFLPs), SSRs (microsatellites), gene-specific genomic probes, and ‘overgo’

Tree Genetics & Genomes

probes derived from BAC end sequences. The total number of markers incorporated into the physical framework was increased to 2,636 markers. Selectively, we fingerprinted all hybridization positive BACs omitted during random fingerprinting. As a result, we incorporated an additional 1× peach genome equivalent composed of marker-positive BACs. Finally, we took advantage of a high-information content fingerprinting (HICF) technique along with improved FPC v8.5.3 software to increase the average number of bands per BAC clone and improve accuracy in contig assembly (Nelson et al. 2007). Here we report an initial HICF physical map for the peach which consists of 2,138 contigs. Of these, 252 contigs are anchored to eight linkage groups of the Prunus reference map. The length of physical map contigs has been estimated at 303 Mb, which is close to the estimated size of the peach genome (Baird et al. 1994). Due to the abundance of hybridization data, the HICF physical map for peach is biased to the expressed genomic regions and thus substantially covers the euchromatic portion of the peach genome.

Materials and methods Sources of BAC clones Two complementary large-insert genomic libraries were used for BAC DNA fingerprinting. ‘The Nemared’ BAC library, constructed in the HindIII site of pBeloBAC11, consists of 44,160 BAC clones with an average insert size of 70 kb giving 8.8-fold genome coverage (Georgi et al. 2002). The second library is based on a haploid strain (Plov 2-1N) derived from the peach rootstock Lovell. The Lovell library was constructed using Sau3A1 fragments cloned into the BamHI site of pIndigoBAC-5. It consists of 34,560 clones with an average insert size of 80 kb providing a 9.2fold genome coverage (L. Georgi, unpublished). An acrylamide gel-based physical map for peach was released in April 2006 (http://www.bioinfo.wsu.edu/gdr/ WebChrom/peach). This map comprises 11,193 BACs assembled into contigs, is integrated with the general Prunus genetic map, and has 1,719 ESTs positioned on it. To improve on this map, these “assembled” BACs were reextracted for HICF together with a set of 9,383 “positive” BACs identified by colony filter hybridizations with 2,636 probes including 228 molecular genetic markers and 2,239 ESTs. After eliminating duplications between the “assembled” and “positive” sets, approximately 12,000 assembled and/or positive BACs were fingerprinted along with 6,000 randomly selected BACs from the haploid Lovell BAC library. Altogether, these HICF fingerprinted BACs represent 4.3× coverage of the peach genome.

The BAC clones were de-condensed from 384- to 96-well format for random fingerprinting or re-ordered into 96-well format for selective fingerprinting using a Q-bot (Genetics, USA). Two sequenced Nemared BACs, 028F08 (Gene Bank #AC154900) and 087G02 (Gene Bank #DQ863257), were incorporated in each plate (positions E7 and H12). The bacterial cultures were manually inoculated from glycerol stocks into 1.2 ml of Terrific Broth–chloramphenicol media (96-deep-well plates) and grown for 21 h at 37°C 250 rpm in a C25 Incubator shaker (New Brunswick Scientific, NJ, USA). BAC DNA isolation in 96-well format followed a standard alkaline lysis miniprep technique (Birnboim and Doly 1979) using AcroPrep96_1ml glass fiber filter plates (Pall Filters, KS, USA). DNA was recovered by isopropanol precipitation followed by washes with 70% ethanol. Typical DNA yield was in the range of 0.7–1.5 μg per BAC clone. Fingerprinting reactions and capillary electrophoresis Fingerprinting reactions were executed using five restriction enzymes (four 6-bp cutters—BamHI, EcoRI, XbaI, and XhoI; one blunt 4-bp cutter—HaeIII) and the ABI PRIZM® SNapShot® restriction fragment labeling kit according to Luo et al. (2003). Briefly, BAC DNA was dissolved overnight in 42 μl of RNase/DNase-free water (Invitrogen). A total of 9.0 μl of a digestion cocktail containing 2.5 U each of BamHI, XbaI, XhoI, and HaeIII and 5 U of EcoRI restriction endonucleases (New England BioLabs, MA, USA); 1× NEbuffer 2 (50 mM NaCl, 10 mM MgCl2, and 1 mM dithiothreitol in 10 mM Tris–HCl, pH 7.9); and 2 μl (1:40 U) RNAse A/T1 cocktail (Ambion, TX, USA) was added to BAC DNA samples. DNA was digested at 37°C for 3 h and transferred into 96-well PCR plates using a Hydra96 semi-automatic dispenser (Robbins Scientific, CA, USA). A SNapShot labeling solution was made of a 1:10 dilution of a SNapShot Multiplex Ready Reaction Mix (ABI, International) in 1× NEbuffer 2. Ten microliters of a labeling solution was transferred to each well and incubated at 65°C for 60 min. Labeled DNA fragments were coldethanol precipitated in presence of 0.25 M sodium acetate (pH 5.2), washed with 70% ethanol twice, and air-dried. Labeled fragments were dissolved in 10 μl of Hi–Di formamide containing 0.15 μl of an internal size standard LIZ-500 and, after a 5-min denaturation at 95°C, resolved on an ABI 3730 automated capillary sequencer using 50-cm capillary arrays filled with POP6 polymer. All chemicals and supplies used for HICF fingerprinting met grade requirements of the instrument’s manufacturer. Data processing Data processing employed a semi-automatic data collection pipeline established in the Clemson University Genomics

Tree Genetics & Genomes

Institute (http://www.genome.clemson.edu). BAC fingerprint profiles including peak areas, peak heights, and fragment sizes were collected by the instrument-implemented program ABI Data Collection v2.0 and automatically scored using the GeneMapper v4.0 software package (Applied Biosystems). An ABI sequencer-compatible package, GenoProfiler v2.0 (You et al. 2007; http://wheat.pw.usda.gov/Physical Mapping), was used for fragment analysis, data filtering from background noise, and removal of repetitive and vector bands. Optimal parameters for background filtering were set up based on fragment frequency analysis for control clone 028F08 fingerprints extracted from the data files and its virtual fingerprint. FPC assembly An HICF-compatible version of FPC (v8.5.3, downloaded from the Arizona Genomics Computational Lab WEB browser, http://www.agcol.arizona.edu/software/fpc) was used to assemble BAC clones into contigs (Soderlund et al. 1997; Soderlund et al. 2000; Nelson and Soderlund 2005). BAC contigs were assembled in a stepwise iterative manner. An initial FPC build was constructed at tolerance of 6 with cut-off 1e-28 followed by the automatic DQer function to minimize the number of false positive merges. The DQer parameters allowed no more than 1% of false positive clones in contigs. Contigs were merged using the FPC implemented function “Ends→Ends” with option “match 2 clones”. A sequential step-by-step increase of the cut-off value from 1e-28 to 1e-15 was applied for automatic contig assembly. FPC assembly was continued semi-manually increasing the cut-offs up to 1e-10. The FPC build was finished using the function “Singles→to Ends”, automatically at cut-offs from 1e-27 to 1e-15 and manually at cut-offs from 1e-14 to 1e-10. At this step, any potential merges not supported by hybridization were rejected. Marker hybridization Hybridizations were performed against the Nemared library spotted on high-density filters. The probes were labeled and hybridized as described previously (Horn et al. 2005). Positive BAC clones were verified and assigned to individual probes by re-hybridization to colony dot blots.

GenoProfiler due to failed fingerprinting or insufficient number of fragments. Thus, a total of 16,895 BACs were submitted for contig assembly using FPCv8.5.3. These clones provide approximately 4.3-fold peach genome coverage (Table 1). Among those clones, 61.1% (10,317 clones), equivalent to 2.5× haploid genomes, were from the HindIII Nemared library and 38.9% (6,578 clones), equivalent to 1.8× haploid genomes, were from the Sau3A1 Lovell library. Fingerprint processing and fragment frequency analysis A cross-platform software application, GenoProfiler, was applied for automated editing of sized fingerprinting profiles generated by the ABI Genetic Analyzers (You et al. 2007). Using a batch-processing module, sized fragment information was extracted from data files introduced from GeneMapper, then the background noise and undesired fragments were removed and fragment size files compatible with FPC software were generated. Fingerprint processing and quality checks followed recommendations in Nelson et al. (2005) and You et al. (2007). Control fingerprints for sequenced Nemared clone 028F08 (#AC154900) were collected into a separate file and the proportion of false positive/false negative fragments (F+/F−) was calculated at different settings. Next, actual fingerprints were compared with virtual fingerprints predicted from “in silico” digestion. Only fragments in the range of 50–500 bp were used for calculations. The average number of scored bands per control BAC 028F08 was 61 and, because the clone size is 66.3 kb, there was approximately one band for every 1.1 kb of sequence. The F+/F− proportion in the control BAC was used to adjust ratios for four channel colors, peak width, threshold value for background noise, and tolerance. Based on the control BAC, the optimized four-channel color ratios were 0.50 for the blue (BamHI) channel, 0.34 for the yellow (XbaI) channel, 0.40 for the red (XhoI) channel, and 0.30 for the green (EcoRI) channel. The peak width was optimized and a setting of ‘less than 15.00’ was selected for filtering of the entire dataset. A tolerance of 0.6, stringent background threshold of 400 units and default

Table 1 Source of BACs fingerprinted for the peach framework physical map

Results

Library name

Cloning site

Insert size (kb)a

BACs in FPC

Fragments per BACa

Genome coverage

Nemared Lovell Total

HindIII BamHI

70 80

10,317 6,578 16,895

53 59 56

2.5× 1.8× 4.3×

BAC fingerprinting Using capillary electrophoresis, we fingerprinted a total of 18,141 BAC clones derived from two complementary BAC libraries. Of these, 1,246 BACs (6.9%) were deleted in

a

Mean values

Tree Genetics & Genomes

settings for “true fragment” threshold were selected as recommended in You et al. (2007). Using optimized parameters, the entire dataset was filtered and fragment frequencies in the fragment size file were analyzed. Vector-derived bands were predicted from virtual digestion of the pBeloBAC11 (http://www.tree.caltech.edu/ protocols/pbelobac11-seq.html) and pIndigoBAC-5 (http:// www.epibio.com/sequences/pIndigo_BamHI.asp) sequences. Following Nelson et al. (2005), the vector band frequencies were used for an additional quality check over the entire dataset. The presence of vector bands in more than 93% of BAC fingerprints indicated a high quality of raw data across the plates. Vector bands were subsequently removed from the dataset, as were 63 high-frequency fragments present in more than 12% of the clones. Finally, BAC fingerprints were selected that had 30–180 fragments per clone. After filtering in GenoProfiler, a total of 956,610 fragments representing 18,141 clones were submitted to FPC (an average of 56 fragments per clone). Given an average insert size of 70 kb, one fragment in FPC database is estimated to represent approximately 1.2 kb. This estimate was close to the calculated value of 1.1 kb per fragment in the control BAC clone. Automatic FPC assembly and semi-manual edition Assembly started from the initial stringent build (first step) then a DQer function was applied to split questionable contigs (second step) followed by an incremental phase of contig joining (third step) and singleton incorporation (fourth step). To control the accuracy of assembly, we took advantage of an acrylamide gel-based physical framework for peach (Zhebentyayeva et al. 2006). The presence of the same set of markers in both maps permitted a manual, contig-by-contig control of HICF assembly from the initial automatic build to finishing the map by adding singletons. The initial assembly at cut-off 1e-28 resulted in 2,389 contigs containing a total of 10,434 clones or 61.8% of the clones submitted to FPC. This proportion of automatically assembled BACs versus singletons was similar to that for other plant genomes successfully reconstructed from BACs using SNapShot™ technique and FPC (http://www. genome.clemson.edu/projects/mimulus/pmap). A DQer function was applied to split contigs with more than 1% questionable clones (Qs). This stringent criterion for the presence of questionable clones in contigs was selected to reduce the number of misplaced clones in contigs. For comparison, in the maize HICF physical map (Nelson et al. 2005) and in two monkey-flower maps (http://www. genome.clemson.edu/projects/mimulus/pmap), DQer re-analyzed contigs with >15% Qs and >10% Qs, respectively. DQer resulted in an initial total of 2,583 contigs. Initial contigs were end-merged at 13 successively lower cut-offs terminating at 1e-18. Each end-merge was con-

firmed by two independent pairs of overlapping BACs to prevent false joints resulting from a single questionable clone in contig. A total of 379 contigs (14.7%) were joined at this step. Next, five semi-manual rounds of merges were performed ending with a cut-off of 1e-15. Potential merges were identified and manually executed using FPC. In the same stepwise-down manner, singletons were integrated into contigs. The assembly was finalized manually at cutoff 1e-10. At this step, contig merges and incorporation of singletons were rejected unless supported by at least three hybridization hits in the resulting contigs. A physical map summary is presented in Table 2. The final build contains 245,048 bands (15,655 clones) assembled into 2,138 contigs and covers 303 Mb or 104.5% of the peach genome (using the estimate of 1.2 kb per fragment). Of these contigs, 384 (or 18%) consist of more than ten clones. The majority of contigs in the database contained three to nine clones (1,412 or 66.0% of the contigs). The accuracy of the BAC contigs was tested using the FPC functions of Calc consensus bands (CB) map, the Contig, and the Fingerprint window (Soderlund et al. 2000). First, the CB maps of fingerprints for each contig were checked twice before and after incorporation of singletons to reveal contigs with Qs having more than 50% unmatched fingerprint bands. Before integrating singletons, only 11 contigs (0.5%) had one or two questionable clones. In the finalized HICF build, only 224 or 10.5% of the contigs had one to four Table 2 Peach physical map summary

Number of clones fingerprinted Number of clones used for contig assembly Number of singletons Number of clones in contigs Number of contigs Size of contigs 400–599 clones 200–399 clones 100–199 clones 50–99 clones 26–49 clones 10–25 clones 3–9 clones 2 clones Number of anchored contigs Physical length of the contigs Physical length of the anchored contigs a

PAAGa map

HICFa map

22,684 18,425

18,141 16,895

8,267 10,128 1,401

1,240 15,655 2,138

0 0 1 0 11 319 820 250 154 197 Mb 33 Mb

1 4 2 4 17 356 1,412 342 252 303.0 Mb 45.0 Mb

PAAG map—an enhanced version of acrylamide gel-based physical map released in April 2006 (http://www.bioinfo.wsu.edu/gdr); HICF map—a physical map based on high-information content fingerprinting using SNapShot labeling kit

Tree Genetics & Genomes Table 3 Markers integrated into physical map for peach (global hybridization data) Hybridization probe

AFLP BAC sequence derived (BAC ends and BAC end overgos) EST (peach unigene set) cDNA probes Gene-specific probes RFLP

SSR

SCAR Total a

Number of probes

17 29 2,239 93 14 190

52

2 2,636

Genetically anchored probes (Prunus genetic maps available at GDR) 11 (SC×B)

127 (T×E) 22 (F×T) 21 (P×F) 4 (J×F) 3 SSRs (T×E) 21 EST-SSRs (bin map) 8 BES-SSRs (bin map) 10 BES-SSR (P×F) 1 (P2175×GN) 228

Positive BACs

Average BACs per probe

151 121

8.9 4.2

7,548 355 135 944

3.4 3.8 9.6 5.0

105

2.8a

24 9,383

12.0 3.6

‘In silico’-assigned BES-SSRs are excluded

questionable clones derived mainly from added singletons. Contig score was another indicator of contig reliability. Contig scores ranging from 0.80 to 1.00 and from 0.70 to 1.00 were accepted for contigs without and with hybridization hits, respectively. Thus, the FPC tests supported the accuracy of assembled BAC contigs. Global hybridization data and map verification Hybridization data were used to integrate developing physical and transcript maps into the HICF map and to further verify the established physical framework. In total, 2,239 ESTs from the peach unigene set (PP_LE) and 93 cDNA probes (peach LF, tomato COS, and sorrel RA_XE probes) were integrated into the HICF physical framework (Table 3). Additionally, BAC end sequence (BES)-derived SSRs (serial names pchgms) and EST-SSRs (serial names EPPCU), both mapped onto the Prunus reference map as well as BAC sequence-based overgo probes, were integrated into the physical framework. Fourteen gene-specific probes were included in hybridizations as well. A total of 2,636 markers were incorporated into the FPC database. Noticeably, a redundancy of 3.6 clones per marker supported the BAC clones redundancy in FPC. Most of the markers (96.0% or 2,530 markers) were positioned on the physical framework (Table 4). Of these markers, 70.0% hit one or two contigs (1,200 and 647 markers, respectively). This statistic strongly supports the accuracy of the contig assembly. As shown in Table 4, in the final HICF build, 683 (25.9%) markers hybridized to multiple contigs. These contigs were inspected manually to determine whether they

were assembled erroneously or if there were ambiguous hybridization scores. CB maps were calculated for contigs with multiple marker hits and the contigs were evaluated for merges using FPC function “Ends→Ends”. Potential contig merges were not found even at low stringency cutoffs from 1e-10 to 1e-04. Therefore, multiple contigs sharing hybridization hits seemingly were not related. Markers assigned to multiple contigs might arise from several reasons such as false positive hybridization signals due to the presence of short repetitive DNA stretches in the probes, duplication of genomic regions, and false contig joints (Klein et al. 2000; Chen et al. 2002; Cone et al. 2002; Yim et al. 2002). Manual editing of conflicting markers/ contig integrations using additional BAC end sequence information could resolve potential false joints in the peach HICF map assembly.

Table 4 Marker statistics (global hybridization data) for the peach physical map

Number of markers used in total One marker in one contig One marker in two contigs One marker in multiple contigs Number of markers in singletons Number of contigs containing markers Number of contigs without markers a

PAAG map April 2006

HICF map May 2007

2,060 (153)a 1,031 283 95 481 813 588

2,636 (159)a 1,200 647 683 106 1,593 545

Genetic markers included in total number of markers

Tree Genetics & Genomes

Anchoring to the Prunus map The initial peach transcriptome map (Horn et al. 2005) used a physical framework consisting of 679 peach BACs hybridizing to 153 peach molecular genetic markers (141 probes). Of these core markers, 79 markers representing 73 probes were mapped on the T×E map by Joobeur et al. (1998). The remaining probes were mapped in crosses other than the reference T×E cross. We replaced these markers with additional markers also mapped in the T×E cross (Dirlewanger et al. 2004; Howad et al. 2005). Thus, the entire framework file of the HICF map now consists of 165 markers mapped in the T×E cross. Of these markers, 133 derived from 127 RFLP probes mapped on the Prunus map. Most of the probes (122) resolved unique anchor points for BAC contig assignment. Five probes resulted in more than one location for anchoring contigs. Also, contig assignment employed 32 mapped, single-copy SSR markers, i.e., SSRs, EST-SSRs, and BES-SSRs. EST-SSRs and BES-SSRs were “in silico” addressed to the BACs based on EST hybridization data or BAC electronic sequence. Therefore, the refined peach physical framework contains 154 (97%) single-copy genetic markers. These 154 markers have the same average spacing of ∽4 cM as reported for the initial transcriptome map (Horn et al. 2005). In the HICF physical Fig. 1 Example of a peach HICF contig anchored to the Prunus reference map. This contig (contig 80) was assigned to the Prunus G8 by RFLP marker AG2 (mapped to G8 at 6.6 cM) and two SSR markers EPPCU3216 and pchgms51 (both mapped to bin 8: 11). The highlighted clones were positive to RFLP probe AG2. Positive clones for SSR EPPCU3216 and EST marker PP_Lea0015E21f are shown in the marker boxes. Four and two hits for EPPCU3216 and PP_Lea0015E21f, respectively, confirmed assembly positive BAC clone in contig 80; one hit for each probe indicated a potential merge with contig 118. Due to co-localization with genetic markers, eight EST markers (PP_LEa probes) from the peach unigene set were positioned on the Prunus genetic map

framework, genetic markers are regularly distributed along the eight linkage groups. The list of framework markers and their positions on the Prunus map are summarized in Table 1 of the Electronic Supplementary Material. An example of a peach BAC contig anchored to the Prunus reference map is shown in Fig. 1. In the current HICF map, 252 contigs or 15.8% of the total hybridized with molecular markers that enabled integration with the reference Prunus genetic map. The physical length of anchored contigs was estimated at 45.0 Mb or 15.5% of the peach genome. Of anchored contigs, 87 contigs representing 15.9 Mb of the peach genome have been accurately positioned and oriented on the Prunus map due to co-localization of core genetic markers or due to multiple hybridization hits from single-copy genetic probes (Table 5). In addition, ten contigs could not be addressed properly to linkage groups because of conflicting marker hybridizations. One half of the contradictory assignments involved genetic markers from Prunus linkage group1 (G1), the most markersaturated chromosome of peach. Most misaligned contigs have relatively low contig scores in CB maps and are likely derived from repetitive and/or centromeric regions of the peach genome. The HICF-based physical map integrated with the peach transcriptome map is publicly available through the GDR

Tree Genetics & Genomes Table 5 Linkage group distribution of anchored and assigned contigs

Chromosome Chromosome Chromosome Chromosome Chromosome Chromosome Chromosome Chromosome Total a

1/G1 2/G2 3/G3 4/G4 5/G5 6/G6 7/G7 8/G8

Genetic markers

Anchored contigs

Assigned contigs

31 24 17 17 19 21 14 16 159

45 48 24 15 35 41 16 18 242

17 14 10 8 12 11 8 8 87

Contigs with conflicting assignment 6 1 1 1 1

(G3, 5, 6, 7)a (G3) (G5) (G5, 8) (G6)

10

Conflicting groups

database (http://www.rosaceae.org/). GDR contains comprehensive data for the genetically anchored peach physical map and annotated EST databases of peach. The web-based programs WebFPC and WebChrom provide a view and display of FPC contigs.

Discussion Overall genome coverage We developed a BAC-based framework physical map of the peach genome using HICF. This map contains 2,138 contigs assembled from 16,895 clones and covers 303 Mb of the peach genome at a redundancy of 4.3× genome equivalents. Physical contigs are anchored to the Prunus genetic maps by 136 core genetic markers and 29 binmapped SSRs regularly spaced along peach chromosomes. The physical length of anchored contigs is estimated to span 15.5% of the genome. In physical maps constructed using traditional fingerprinting on agarose or acrylamide gels, the total length of assembled contigs was estimated to exceed the size of corresponding genomes by 32% in chicken (Ren et al. 2003), by 23.6% in apple (Han et al. 2007), and by 26.3% in soybean (Wu et al. 2004). An excessive physical map length usually indicates that most contigs overlap adjacent contigs, although overlaps were not detected under conditions used for fingerprinting. Manual editing of initial contigs with support from hybridization data usually shrinks the physical map length down to an “estimated genome size ±5%”, as reported for thale cress (Mozo et al. 1999) and for rice (Chen et al. 2002). In the HICF map reported here, the physical length was estimated at 104.5% of the peach genome. Commensurate estimates for genome coverage were obtained in other genome-wide physical mapping projects that utilized the HICF technique. In complete maps established for channel

catfish (Quiniou et al. 2007) and monkey flower (http:// www.genome.clemson.edu/projects/mimulus/pmap), the physical length of assembly was close to the estimated sizes of the corresponding genomes. In a simulation study by Nelson et al. (2007), the HICF five-enzyme SNapShot method performed significantly better than other fingerprinting techniques in identifying overlapping clones and produced more compact maps. Therefore, we believe that the 104% estimated coverage of our peach HICF map reflects the superior performance of the HICF technique. We compared FPC statistics from the HICF map assembly for peach with those from the progressive genome assembly for catfish (Quiniou et al. 2007). Contig size distribution in the peach HICF map was similar to that in the catfish assembly at 4× genome coverage. However, the estimated total length of peach HICF contigs was significantly higher than in catfish (104.5% and 73.3%, respectively). Apparently, non-random BAC fingerprinting contributed to a broader assembly of the peach genome at a nominal 4.3× map redundancy. By selecting BACs from existing contigs in our acrylamide-based map, the HICF fingerprinting omitted clones affecting calculated genome coverage in random fingerprinting, i.e., chimerical clones, well-to-well contaminated clones, low-information content clones, etc. Furthermore, selective fingerprinting of positive BACs representing ∽1× peach genome equivalent enhanced the efficiency of genome assembly in combination with the supporting hybridization data. Integration of the peach transcript database into the physical map provided another independent evaluation of genome coverage for our HICF build. More than 96% of peach unigene ESTs were positioned on the physical framework during the FPC assembly. Thus, marker statistics of HICF assembly supported the percent genome coverage calculated from FPC summary. Notably, the map redundancy estimated from the fingerprinting statistics was in agreement with that estimated from the global hybridization data. As shown in Table 3, more than 55% of BAC

Tree Genetics & Genomes

clones in our HICF map had hybridization hits (mainly ESTs). The number of contigs with hybridization hits accounted for 74.5% of the total. Due to selective fingerprinting of the EST positive clones, our HICF map is biased to the euchromatic portion of the peach genome. Given that the physical contigs are calculated to span 104.5% of genome, we presume that at least the entire euchromatic portion of the peach genome is represented in our HICF map. Efficiency of HICF fingerprinting Two fingerprinting techniques were used in physical mapping the peach genome. First, we released an initial acrylamide gel-based physical framework (Zhebentyayeva et al. 2006). Subsequently, we used the enhanced version of this framework (http://www.bioinfo.wsu.edu/gdr/ WebChrom/peach/) for verification of an HICF assembly. Thus, our project provides a unique opportunity to compare these two techniques in genome-wide physical mapping. Recently, the efficacy of FPC contig assembly from HICF fingerprints became the center of a fierce debate (Xu et al. 2004; Nelson et al. 2007). Our experience in the application of acrylamide gel-based and HICF (SNapShot, four labeled restriction sites) techniques supported the conclusion of Nelson et al. (2007) that FPC assembly from HICF fingerprints is more efficacious than using radioactively labeled fingerprints on acrylamide gels (one labeled restriction site). Application of the HICF technique to the same BAC libraries resulted in a significant increase of information content in fingerprints from 20 to 56 fragments per clone. The increased number of fragments available for assembly resulted in greater BAC integration into contigs. Two numerical parameters of the HICF build, the number of singletons and the physical length of contigs, reflect this fact (Table 2). The proportion of singletons in the HICF build (7.3% or 1,240 clones) was one sixth that of the acrylamide build (44.9% or 8,267 clones). Simultaneously, the physical length of HICF contigs increased 35.0%, from 197 to 303 Mb, to be commensurate with the predicted size of the peach genome. Thus, in case of BAC libraries with relatively short insert sizes such as peach Nemared and haploid Lovell (with an average insert size 70–80 kb), the HICF technique outperformed acrylamide gel-based fingerprinting. Towards a complete physical map for peach Finalizing the HICF assembly demands several approaches to collapse the developed framework map. Below we discuss the intermediate status of the present framework physical map and describe our current approaches toward the complete physical assembly that supports shotgun sequencing and assembly of the peach genome.

Map redundancy Sevenfold to tenfold genome representations are empirically considered to be a minimal requirement for robust genome-wide assembly of BAC fingerprints using FPC (Meyers et al. 2004; Xu et al. 2004). Based on theoretical analysis, Barillot et al. (1991) proposed a combination of restriction digest fingerprinting and hybridizations as an optimal physical mapping strategy. By combining both mapping techniques, they optimized library redundancy at 7× genome equivalents for accurate contig assembly. Following their observations on progress in contig assembly, we assumed that increasing BAC redundancy up to seven peach genome equivalents is sufficient to collapse HICF framework into a complete physical map. To support further integration of existing contigs into mega-contigs, we developed a new BAC library derived from a di-haploid Lovell genotype with an average insert size of 100 kb (unpublished). Therefore, additional fingerprinting of 8,000–10,000 clones equal to approximately three peach haploid genomes will adjust the HICF map density up to the desired level. Development of sequence-tagged connectors randomly distributed across the genome is another approach for contig expansion and identification of overlapping contigs. Using FPC, we generated a minimum tiling path across the HICF map and selected ~2,200 clones for BES. BES-based overgo hybridizations will not only facilitate collapsing of the HICF framework map into mega-contigs but also will provide landmarks for assembly in the upcoming peach genome sequencing project. Floating contigs In the present map, more than 80% of the peach genome is assembled into floating contigs, i.e., contigs not assigned to any chromosomal location. Most floating contigs (~70%) contain BACs that hybridized to at least one probe included in the global hybridizations in Table 3. In total, 257 Mb (85.0%) of the peach genome is not aligned with Prunus linkage groups/chromosomes due to a shortage of bridging anchor points between the physical and genetic maps. Peach unigene ESTs are the most abundant type of markers incorporated into the FPC database. Positioning ESTs on the physical framework provides a direct link from the annotated functional EST database to genes of interest organized into a contiguous framework on the physical map. Altogether, 1,012 ESTs were anchored to the Prunus map by sharing BACs or contigs with core genetic markers derived from the Prunus genetic map. However, more than half of peach unigene ESTs in FPC database (1,227 ESTs) were positioned in floating contigs and remain to be anchored on the Prunus map. Thus, the Prunus EST database could be an important source for further saturation of the physical framework with genetic markers. To increase the number of anchor points for contig assignment, we exploit the Prunus unigene EST

Tree Genetics & Genomes

collection initially excluded from hybridizations because they contained microsatellite (SSR) repeats. A set of 180 EST-SSRs was identified in EST collections from the GDR (EPPCU and EPDCU series) and Institut National de la Recherche Agronomique-Bordeaux (EPPB series), of which 128 EST-SSRs were positioned on the Prunus map (Dirlewanger et al. 2004; Howad et al. 2005). Using overgo primers designed outside of the SSR repeats, we initiated hybridizations against both Nemared and Lovell libraries, two sources of clones in our physical map. These genetically mapped EST-SSR markers are potential anchor points for positioning contigs on the integrated physical/ genetic framework. Mapped EST-SSR markers will be employed for manual map revision and editing.

mond, and the genomic region responsible for the “evergrowing” mutation in peach. Furthermore, a minimal tile path over physical contigs, especially enhanced with sequenced BAC ends, will facilitate sequence assembly in the upcoming peach genome sequencing project. We believe that the di-haploid peach genotype selected for shotgun sequencing of the peach genome will dramatically increase the effectiveness and accuracy of sequence assembly due to the lack of allelic variation. In the future, a sequenced and assembled genome of peach will dramatically facilitate gene discoveries aimed at improvement of agronomic performance of rosaceous crops and fundamental evolutionary studies across the entire Rosaceae.

Contradictory contig assignments From the lesson of the Arabidopsis genome study, at least 10–15% of nuclear DNA is highly methylated in heterochromatic regions, i.e., cytogenetically defined centromeres, telomeres, and nucleolar organizers (Goodman et al. 1995). Large contigs derived from repetitive genomic regions are reported in most firstgeneration physical maps published to date, for instance, in rice (Tao et al. 2001; Chen et al. 2002), soybean (Wu et al. 2004), sorghum (Klein et al. 2000), apple (Han et al. 2007), and monkey flower (http://www.genome.clemson.edu/ projects/mimulus/pmap). Moreover, in most physical maps, the mitochondrial and chloroplast genomes are organized into large contigs similar to that shown in the Arabidopsis thaliana physical map (Mozo et al. 1999). For manual revision of the peach framework physical map, we identified 11 large contigs each composed of more than 50 BACs (Table 2). The abundance of peach unigene EST marker hits indicated that nine of them (contigs 23, 38, 88, 160, 217, 218, 259, 1,980, and 2,062) were derived from peach genomic DNA. These contigs had relatively low contig scores, a signature of highly repetitive genomic regions enriched with transposable elements, and could not be assigned to specific linkage groups due to contradictory genetic marker hybridizations. The dramatic shortage of EST hybridization hits in the remaining large contigs indicated that they most likely derive from non-nuclear DNA (i.e., plastid and mitochondrial). Additional information such as hybridizations with specific probes is needed to resolve these contradictory contigs and properly assign them to genetic positions on the Prunus map. The framework physical map for peach reported here is a valuable resource for identifying genes of interest, mapbased positional cloning in Prunus species, and comparative sequence analysis across rosaceous species. Comparative sequence analysis is currently underway for several genomic regions involved in important agronomic traits such as resistance to plum pox virus in apricot, root-knot nematode resistance in peach, self-incompatibility in al-

Acknowledgement This project was supported by the United States Department of Agriculture NRI Award # 2005-35300-15452.

References Abbott AG, Georgi L, Inigo M, Sosinski B, Yvergniaux D, Wang Y, Blenda A, Reighard G (2002) Peach: the model genome for Rosaceae. Acta Hort 575:145–155 Abbott AG, Arús P, Scorza R (2006) Peach. In: Kole C (ed) Genome mapping and molecular breeding in plants. Springer, Berlin, pp 137–156 Aranzana MJ, Pineda A, Cosson P, Dirlewanger E, Ascasibar J, Cipriani G, Ryder CD, Testolin R, Abbott A, King GJ, Iezzoni AF, Arús P (2003) A set of simple-sequence repeat (SSR) markers covering the Prunus genome. Theor Appl Genet 106:819–825 Arús P, Yamamoto T, Dirlewanger E, Abbott AG (2006) Synteny in the Rosaceae. In: Janick J (ed) Plant Breeding Reviews vol. 27. Wiley, Hoboken, pp 175–211 Baird WV, Estager AS, Wells J (1994) Estimating nuclear DNA content in peach and related diploid species using laser flow cytometry and DNA hybridization. J Am Soc Hort Sci 119:1312–1316 Barillot E, Dausset J, Cohen D (1991) Theoretical analysis of a physical mapping strategy using random single-copy landmarks. Proc Natl Acad Sci USA 88:3917–3921 Birnboim HC, Doly J (1979) A rapid alkaline extraction procedure for screening recombinant plasmid DNA. Nucleic Acids Res 7:1513–1523 Chen M, Presting G, Barbazuk W, Goicoechea J, Blackmon B, Fang G, Kim H, Frisch D, Yu Y, Higingbottom S, Phimphilai J, Phimphilai D, Thurmond S, Gaudette B, Li P, Liu J, Hatfield J, Sun S, Farrar K, Henderson C, Barnett L, Costa R, Williams B, Walser S, Atkins M, Hall C, Bancroft I, Salse J, Regad F, Mohapatra T, Singh N, Tyagi A, Soderlund C, Dean R, Wing R (2002) An integrated physical and genetic map of the rice genome. Plant Cell 14:537–545 Cone KC, McMullen MD, Bi IV, Davis GL, Yim Y, Gardiner JM, Polacco ML, Sanchez-Villeda H, Fang Z, Schroeder SG, Havermann SA, Bowers JE, Paterson AH, Soderlund CA, Engler FW, Wing RA, Coe EH (2002) Genetic, physical, and informatics resources for maize. On the road to an integrated map. Plant Physiol 130:1598–1605 Dirlewanger E, Graziano E, Joobeur T, Garriga-Caldere F, Cosson P, Howad W, Arús P (2004) Comparative mapping and marker-

Tree Genetics & Genomes assisted selection in Rosaceae fruit crops. Proc Natl Acad Sci USA 101:9891–9896 Dirlewanger E, Cosson P, Boudehri K, Renaud C, Capdeville G, Tauzin Y, Laigret F, Moing A (2006) Development of a secondgeneration genetic linkage map for peach [Prunus persica (L.) Batsch] and characterization of morphological traits affecting flower and fruit. Tree Genetics & Genomes 3:1–13 Draye X, Lin YR, Qian XY, Bowers JE, Burow GB, Morrell PL, Peterson DG, Presting GG, Ren SX, Wing RA, Paterson AH (2001) Toward integration of comparative genetic, physical, diversity, and cytomolecular maps for grasses and grains, using the sorghum genome as a foundation. Plant Physiol 125:1325–1341 Evans RC, Campbell CS (2002) The origin of the apple subfamily (Maloideae; Rosaceae) as clarified by DNA sequence data from duplicated GBSSI genes. Am J Bot 89:1478–1484 Georgi LL, Wang Y, Yverggniaux D, Ormsbee T, Inigo M, Reighard GL, Abbott AG (2002) Construction of a BAC library and its application to the identification of simple sequence repeats in peach [Prunus persica (L.) Batsch]. Theor Appl Genet 105:1151–1158 Gladkova VN (1972) On the origin of subfamily Maloideae. Bot Zhurnal 57:42–49 (in Russian) Goodman HM, Ecker JR, Dean C (1995) The genome of Arabidopsis thaliana. Proc Natl Acad Sci USA 92:10831–10835 Green ED (2001) Strategies for the systematic sequencing of complex genomes. Natl Rev Genet 2:573–583 Han J, Gasić K, Marron B, Beever JE, Korban SS (2007) A BACbased physical map of the apple genome. Genomics 89:630–637 Hoskins RA, Nelson CR, Berman BP, Laverty TR, George L, Ciesiolka L, Naemuddin M, Arenson AD, Durbin J, David RG, Tabor PE, Bailey MR, DeShazo DR, Catanese J, Mammoser A, Osoegawa K, de Jong PE, Celniker SE, Gibbs RA, Rubin GM, Scherer SE (2000) BAC-based physical map of the major autosomes of Drosophila melanogaster. Science 287:2271–2274 Howad W, Yamamoto T, Dirlewanger E, Testolin R, Cosson P, Cipriani G, Monforte AJ, Georgi L, Abbott AG, Arús P (2005) Mapping with a few plants: using selective mapping for microsatellite saturation of the Prunus reference map. Genetics 171:1305–1309 Horn R, Lecouls A-C, Callahan A, Dandekar A, Garay L, McCord P, Howad W, Chan H, Verde I, Ramaswamy K, Main D, Jung S, Georgi L, Forrest S, Mook J, Zhebentyayeva TN, Yu Y, Kim HR, Jesudurai C, Sosinski BA, Arús P, Baird V, Parfitt D, Reighard G, Scorza R, Tomkins J, Wing R, Abbott AG (2005) Candidate gene database and transcript map for peach, a model species for fruit trees. Theor Appl Genet 110:1419–1428 Joobeur T, Viruel MA, de Vicente MC, Jauregui B, Ballester J, Dettori MT, Verde I, Truco MJ, Messeguer R, Batlle I (1998) Construction of a saturated linkage map for Prunus using an almond×peach F2 progeny. Theor Appl Genet 97:1034–1041 Jung S, Jesudurai C, Staton M, Du Z, Ficklin S, Cho I, Abbott A, Tomkins J, Main D (2004) GDR (Genome Database for Rosaceae): integrated web resources for Rosaceae genomics and genetics research. BMC Bioinformatics 5:130 Jung S, Main D, Staton M, Cho I, Zhebentyayeva T, Arus P, Abbott A (2006) Synteny conservation between the Prunus genome and both the present and ancestral Arabidopsis genomes. BMC Genomics 7:81 Klein PE, Klein RR, Cartinhour SW, Ulanch PE, Dong J, Obert JA, Morishige DT, Schlueter SD, Childs KL, Ale M, Mullet JE (2000) A high-throughput AFLP-based method for constructing integrated genetic and physical maps: progress toward a sorghum genome map. Genome Res 10:789–807 Lalli DA, Decroocq V, Blenda AV, Schurdi-Levraud V, Garay L, Le Gall O, Damsteegt V, Reighard GL, Abbott AG (2005) Identification and mapping of resistance gene analogs (RGAs)

in Prunus: a resistance map for Prunus. Theor Appl Genet 111:1504–1513 Luo MC, Thomas C, You FM, Hsiao J, Ouyang S, Buell CR, Malandro M, McGuire PE, Anderson OD, Dvorak J (2003) Highthroughput fingerprinting of bacterial artificial chromosomes using the snapshot labeling kit and sizing of restriction fragments by capillary electrophoresis. Genomics 82:378–389 Marra M, Kukaba T, Sekhon T, Hillier L, Martiensen R, Chinwalla A, Crokett J, Fedele J, Grover H, Gund G, McCombie WR, McDonald K, McPherson J, Mudd N, Parnell L, Schein J, Seim R, Shelby P, Waterson R, Wilson R (1999) A map for sequence analysis of the Arabidopsis thaliana genome. Nature Genet 22:269–270 Meyers BC, Scalabrin S, Morgante M (2004) Mapping and sequencing complex genomes: let’s get physical. Nat Rev Genet 5:578–588 Mozo T, Dewar K, Dunn P, Ecker JR, Fischer S, Kloska S, Lehrach H, Marra M, Martienssen R, Meier-Ewert S, Altmann T (1999) A complete BAC-based physical map of the Arabidopsis thaliana genome. Nat Genet 3:271–275 Nelson WM, Bharti AK, Butler E, Wei F, Fuks G, Kim HR, Wing RA, Messing J, Soderlund K (2005) Whole-genome validation of high-information-content fingerprinting. Plant Physiol 139:27–38 Nelson WM, Soderlund C (2005) Software for restriction fragment physical maps. In: Meksem K, Kahl G (eds) The handbook of plant genome mapping: genetic and physical mapping. WileyVCH, Weinheim, pp 285–306 Nelson WM, Dvorak J, Luo MC, Messing J, Wing RA, Soderlund C (2007) Efficacy of clone fingerprinting methodologies. Genomics 89:160–165 Potter D (2003) Molecular phylogenetic studies in Rosaceae. In: Sharma AK, Sharma A (eds) Plant genome: biodiversity and evolution, vol. 1, part A. Science Publishers, Enfield, pp 319–351 Quiniou SM-A, Waldbieser GC, Duke MV (2007) A first generation BAC-based physical map of the channel catfish genome. BMC Genomics 8:40 Ren C, Lee MK, Yan B, Ding K, Cox B, Romanov MN, Price JA, Dodgson JB, Zhang HB (2003) A BAC-based physical map of the chicken genome. Genome Res 13:2754–2758 Sánchez-Pérez R, Howad W, Dicenta F, Arús P, Martínez-Gómez P (2007) Mapping major genes and quantitative trait loci controlling agronomic traits in almond. Plant Breeding 126:310–319 Soderlund C, Longden I, Mott R (1997) FPC: a system for building contigs from restriction fingerprinted clones. CABIOS 13: 523–535 Soderlund C, Humphray S, Dunham A, French L (2000) Contigs built with fingerprint, markers, and FPC V4.7. Genome Res 10: 1772–1778 Takhtajan A (1997) Diversity and classification of flowering plants. Columbia University Press, Columbia, p 643 Tao Q, Chang Y-L, Wang J, Chena H, Islam-Faridi MN, Scheuring C, Wang B, Stelly DM, Hong-Bin Zhang H-B (2001) Bacterial artificial chromosome-based physical map of the rice genome constructed by restriction fingerprint analysis. Genetics 158:1711–1724 Wu C, Sun S, Nimmakayala P, Santos FA, Meksem K, Springman R, Ding K, Lightfoot DA, Zhang H-B (2004) A BAC- and BIBACbased physical map of the soybean genome. Genome Res 14:319–326 Xu Z, Sun S, Covaleda L, Ding K, Zhang A, Wu C, Scheuring C, Zhang HB (2004) Genome physical mapping with large-insert bacterial clones by fingerprint analysis: methodologies, source clone genome coverage, and contig map quality. Genomics 84:941–951 Yim YS, Davis GL, Duru NA, Musket TA, Linton EW, Messing JW, McMullen MD, Soderlund CA, Polacco ML, Gardiner JM, Coe EH Jr (2002) Characterization of three maize

Tree Genetics & Genomes bacterial artificial chromosome libraries toward anchoring of the physical map to the genetic map using high-density bacterial artificial chromosome filter hybridization. Plant Physiol 130:1686–1696 You FM, Luo M-C, Gu YQ, Lazo GR, Deal K, Dvorak J, Anderson OD (2007) GenoProfiler: batch processing of high throughput capillary fingerprinting data. Bioinformatics 23:240–242

Zhang HB, Wing RA (1997) Physical mapping of the rice genome with BACs. Plant Mol Biol 35:115–127 Zhang HB, Wu C (2001) BAC as tools for genome sequencing. Plant Physiol Biochem 39:195–209 Zhebentyayeva TN, Horn R, Mook J, Lecouls A-C, Georgi L, SwireClark G, Reighard GL, Baird WV, Abbott AG (2006) A physical framework for the peach genome. Acta Hort 713:83–88