Genomic organization, gene duplication, and expression analysis of interleukin-1 beta in channel catfish (Ictalurus punctatus)

Interleukin-1 beta (IL-1 beta) is one of the pivotal early response pro-inflammatory cytokines that enables organisms to respond to infection and induces a cascade of reactions leading to inflammation. In spite of its importance and two decades of studies in the mammalian species, genes encoding IL-1 beta were not identified from non-mammalian species until recently. Recent research, particularly with genomic approaches, has led to sequencing of IL-1 beta from many species. Clinical studies also Suggested IL-1 beta as an immunoreagulatory molecule potentially useful for enhancing vaccination. However, no IL-1 beta genes have been identified from channel catfish, the primary aquaculture species from the United States. In this study, we identified two distinct cDNAs encoding catfish IL-1 beta. Their encoding genes were identified, sequenced, and characterized. The catfish IL-1 beta genes were assigned to bacterial artificial chromosome (BAC) clones. Genomic studies indicated that the IL-1 beta genes were tandemly duplicated on the same chromosome. Phylogenetic analysis of various IL-1 beta genes indicated the possibility of recent species-specific gene duplications in channel catfish, and perhaps also in swine and carp. Expression analysis indicated that both IL-1 beta genes were expressed, but exhibited distinct expression profiles in various catfish tissues, and after bacterial infection with Edwardsiella ictaluri. (c) 2005 Elsevier Ltd. All rights reserved.

Chromosomal in situ hybridization of Cyprinus carpio genomic DNA repetitive sequence CR1

Common carp Cyprinus carpio genomic DNA repetitive sequence CR1 has been DIG-labeled and hybridized in situ against chromosomes of red common carp (Cyprinus carpio L. Xingguo red var.). It is found that the repetitive sequence CR1 is mainly localized at the centromeric regions of chromosomes of the red common carp, The application of the chromosomal in situ hybridization technique on fish and the relationship between CR1 repetitive sequence distribution and its function have been discussed.

簇毛麦（ Dasypyrum villosum）矮秆突变体的研究

株高是农作物的重要农艺性状之一，适度矮化有利于农作物的耐肥、抗倒、高产等。20世纪50年代，以日本的赤小麦为矮源的半矮秆小麦的培育和推广，使得世界粮食产量显著增长，被誉为“绿色革命”。迄今为止，已报到的麦类矮秆、半矮秆基因已达70多个，但由于某些矮源极度矮化或者矮化的同时伴随不利的农艺性状，使得真正运用于育种实践的矮源较少。因此，发掘和鉴定新的控制麦类作物株高的基因，开展株高基因定位、克隆及作用机理等方面的研究，对实现麦类作物株高的定向改良，具有重要的理论意义和应用价值。簇毛麦(Dasypyrum villosum，2n＝14，VV)是禾本科簇毛麦属一年生二倍体异花授粉植物，为栽培小麦的近缘属。本课题组在不同来源的簇毛麦杂交后代中发现了一株自然突变产生的矮秆突变体。观察分析了该突变体的生物学特性，对矮秆性状进行了遗传分析，对茎节细胞长度、花粉的活力进行了细胞学观察，考察了该突变体内源赤霉素含量及不同浓度外施赤霉素对突变体的作用，分析了赤霉素生物合成途径中的内根贝壳杉烯氧化酶（KO）和赤霉素20氧化酶(GA20ox)的转录水平，对赤霉素20氧化酶和赤霉素3-β羟化酶(GA3ox)进行了克隆和序列分析，并对GA20ox进行了原核表达和表达的组织特异性研究。主要研究结果如下：1. 该突变体与对照植株在苗期无差异，在拔节后期才表现出植株矮小，相对对照植株，节间伸长明显受到抑制，叶鞘长度基本不变。在成熟期，对照植株的平均株高为110cm，而突变株的平均株高为32cm，仅为对照植株的1/3 左右。除了株高变矮以外，在成熟后期，突变株还表现一定程度的早衰和雄性不育。I2-KI染色法观察花粉活力结果表明，对照植株花粉90%以上都是有活力的，而突变植株的花粉仅20%左右有活力。2. 突变株与对照植株的杂交F1代均表现正常株高，表明该突变性状为隐性突变。F1代植株相互授粉得到的168株F2代植株中，株高出现分离，正常株高（株高高于80cm）与矮秆植株（株高矮于40cm）的株数比为130：38，经卡方检验，其分离比符合3：1的分离比，因此推测该突变体属于单基因的隐性突变。3. 用ELISA方法检测突变株和对照植株的幼嫩种子中内源性生物活性赤霉素（GA1＋3）含量，结果表明突变株的赤霉素含量为36 ng/ml，而对照植株的赤霉素含量为900 ng/ml。对突变株外施赤霉素，发现矮秆突变株的株高和花粉育性均可得到恢复。这些结果表明该突变株为赤霉素缺陷型突变。4. 用荧光定量PCR方法比较突变株与对照植株中内根贝壳杉烯氧化酶和赤霉素20氧化酶的转录水平，结果表明突变株的KO转录水平比对照植株分别提高了6倍（苗期）和16倍（成熟期），突变株的GA20ox转录水平与对照植株在苗期无明显差异，在成熟期突变株较对照植株则提高了10倍左右。这些结果表明该矮秆突变体与赤霉素的生物合成途径密切相关，而且极有可能在赤霉素的生物合成途径早期就发生了改变。5. 以簇毛麦总基因组为模板，同源克隆了GenBank登录号为EU142950，RT-PCR分离克隆了簇毛麦的GA3ox基因cDNA全长序列，分析结果表明该cDNA全长1206bp，含完整编码区1104bp，推测该序列编码蛋白含368个氨基酸残基，分子量为40.063KD，等电点为6.27。预测的氨基酸序列含有双加氧酶的活性结构，在酶活性中心2个Fe离子结合的氨基酸残基非常保守。该序列与小麦、大麦和水稻的GA3ox基因一致性分别为98%、96%、86%。基因组序列与cDNA序列在外显子部分一致，在478-715bp和879-1019bp处分别含238bp和140bp的内含子。6. 通过RT-PCR技术克隆了簇毛麦的GA20ox基因全长，命名为DvGA20ox，GenBank登录号为EU142949。该基因全长1080个碱基，编码359个氨基酸，具有典型的植物GA20ox基因结构。该基因编码的蛋白质与小麦、大麦、黑麦草等GA20ox蛋白的同源性分别为98%，97% 和91%。该序列重组到原核表达载体pET-32a(+)上，将获得的重组子pET-32a(+)-DvGA20ox转化大肠杆菌BL21pLysS后用IPTG进行诱导表达。SDS-PAGE分析表明，DvGA20ox基因在大肠杆菌中获得了高效表达，融合蛋白分子量为55kDa。定量PCR分析表明，该基因在簇毛麦不同器官中的表达差异明显：叶片中表达水平最高，根部表达水平次之，茎部和穗中表达较弱。在外施赤霉素后，该基因的表达水平在两小时以后急剧下降，表明该基因的表达受自身的反馈调节。本研究结果认为，(1)该簇毛麦矮秆突变体为单基因的隐性突变；(2)该矮秆突变体为赤霉素敏感突变，内源赤霉素含量检测表明突变体的内源性赤霉素含量仅为对照植株的1/30；(3)荧光定量PCR结果表明突变株的赤霉素生物合成途径的关键酶基因表达水平比对照植株高，而且突变植株的赤霉素生物合成改变很可能发生在赤霉素生物合成途径的早期；(4)GA20ox有表达的组织特异性，且受到自身产物的反馈调节。 Plant height is an impotrant agronomic trait of triticeae crops.Semi-dwarf cropcultivars, including those of wheat, maize and rice, have significantly increased grainproduction that has been known as “green revolution”. The new dwarf varieties couldraise the harvest Index at the expense of straw biomass, and, at the sametime, improvelodging resistance and responsiveness to nitrogen fertilizer. Moreover, dwarf traits ofplant are crucial for elucidating mechanisms for plant growth and development aswell. In many plant species, various dwarf mutants have been isolated and theirmodles of inheritance and physiology also have been widely investigated.The causesfor their dwarf phenotypes were found to be associated with plant hormones,especially, gibberellins GAs.Dasypyrum villosum Candargy (syn.Haynaldia villosa) is a cross-pollinating,diploid (2n = 2x = 14) annual species that belongs to the tribe Triticeae. It is native toSouthern Europe and West Asia, especially the Caucasuses, and grows underconditions unfavorable to most cultivated crops. The genome of D. villosum,designated V by Sears, is considered an important donor of genes to wheat for improving powdery mildew resistance, take-all, eyespot, and plant and seed storageprotein content. A spontaneous dwarf mutant was found in D. villosum populations.The biological character and modles of inheritance of this dwarf mutant are studied.The cell length of stem cell is observed. The influence of extraneous gibberellin tothe dwarf mutant is also examined; the transcript level of key enzyme of gibberellinbiosynthesis pathway in mutant and control plants is compared. GA3ox and GA20oxare cloned and its expression pattern is researched.1. The dwarf mutant showed no difference with control plants at seedlingstage.At mature stage, the average height of control plants were 110cm and the dwarfplants were 33cm. The height of the mutant plant was only one third of the normalplants due to the shortened internodes. Cytology observation showed that theelongation of stem epidermal and the parenchyma cells were reduced. The dwarfmutant also shows partly male sterile. Pollen viability test indicates that more than80% of the pollen of the mutant is not viable.2. The inheritance modle of this dwarf mutant is studied. All The F1 plantsshowed normal phenotype indicating that the dwarfism is controlled by recessivealleles. Among the 168 F2 plants, there are 130 normal plants and 30 dwarf plants, thesegregation proportion accord with Mendel’s 3:1 segregation. We therefore proposethat this dwarf phenotype is controlled by a single recessive gene.3. Quantitative analyses of endogenous GA1+3 in the young seeds indicated thatthe content of GA1+3 was 36ng/ml in mutant plants and 900ng/ml in normal plants.The endogenous bioactive GA1+3 in mutant plants are only about 1/30 of that innormal plants. In addition, exogenously supplied GA3 could considerably restore themutant plant to normal phenotype. These results showed that this mutant wasdefective in the GA biosynthesis.4. More than ten enzymes are involved in GA biosynthesis. KO catalyzes thefirst cytochrome P450-mediated step in the gibberellin biosynthetic pathway and themutant of KO lead to a gibberellin-responsive dwarf mutant. GA20ox catalyze therate-limited steps so that their transcript level will influence the endogenous GAbiosynthesis and modifies plant architecture. The relative expression levels of genesencoding KO and GA20ox were quantified by real time PCR to assess whether thechanges in GA content correlated with the expression of GA metabolism genes andwhere the mutant occurred during the GA biosynthesis pathway. In mutant plants,the transcript levels of KO increased about 6-fold and 16-fold at the seedling stage and elongating stage respectively comparing with the normal plants. For theseedlings, there was no notable difference in the expression of GA20ox betweenmutant and normal plants. At the elongating stage, GA20ox transcript increased 10times in mutant plants, suggesting that the GA biosynthesis pathway in mutant plantshad changed from the early steps rather than the late steps.5. A full length cDNA of D. villosum gibberellin 3β-hydroxylase homology(designated as DvGA3ox) was isolated and consisted of 1206bp containing an openreading frame of 1104bp encoding 368 predicted amino acid residues. Identityanalysis showed that the gibberellin 3β-hydroxylase nucleotide sequence shared 98%,96% and 86% homology with that of wheat, barley and rice. The predicted peptidecontained the active-site Fe of known gibberellin 3β-hydroxylase and the regionhomologous to wheat, barley and Arabidopsis. The genomic clone of gibberellin3β-hydroxylase has two introns.6. The full-length cDNA of D. villosum gibberellin 20 oxidase (designated asDvGA20ox) was isolated and consisted of 1080-bp and encoded 359 amino acidresidues with a calculated mol wt of 42.46 KD. Comparative and bio-informaticsanalyses revealed that DvGA20ox had close similarity with GA20ox from otherspecies and contained a conserved LPWKET and NYYPXCQKP regions. Tissueexpression pattern analysis revealed DvGA20ox expressed in all the tissues that wereexamined and the highest expression of DvGA20ox in expanding leaves followed byroots. Heterologous expression of this cDNA clone in Escherichia coli gave a fusionprotein that about 55KD. Transcript levels of DvGA20ox dramatically reduced twohours after application of biologically active GA3, suggesting that the biosynthesis ofthis enzymes might be under feedback control.

potential mechanisms involved in resistant phenotype of MCF-7breast carcinoma cells to ionizing radiation induced apoptosis

In the present study, we investigated the mechanisms of apoptosis resistance and the roles of the phosphorylation of BRCA1, p21, the Bax/Bcl-2 protein ratio and cell cycle arrest in IR-induced apoptosis in MCF-7 cells. X-irradiation, in particular at low dose (1 Gy), but not carbon ion irradiation, had a significant antiproliferative effect on the growth of MCF-7 cells. 1 Gy X-irradiation resulted in G1 and G2 phase arrest, but 4 Gy induced a significant G1 block. In contrast, carbon ion irradiation resulted in a significant accumulation in the G2 phase. Concomitant with the phosphorylation of H2AX induced by DNA damage,carbon ion irradiation resulted in an approximately 1.9–2.8-fold increase in the phosphorylation of BRCA1 on serine residue 1524, significantly greater than that detected for X-irradiation. Carbon ion irradiation caused a dramatic increase in p21 expression and drastic decrease in Bax expression compared with X-irradiation. The data implicated that phosphorylation of BRCA1 on serine residue 1524 might,at least partially, induce p21 expression but repress Bax expression. Together, our results suggested that the phosphorylation of BRCA1 at Ser-1524 might contribute to the G2 phase arrest and might be an upstream signal involved in preventing apoptosis signal via upregulation of p21 and downregulation of the Bax/Bcl-2 ratio.

Genomic organization, nucleotide sequence analysis of the core histone genes cluster in Chlamys farreri and molecular evolution assessment of the H2A and H2B

This work represents the nucleotide sequence of the core histone gene cluster from scallop Chlamys farreri. The tandemly repeated unit of 5671 bp containing a copy of the four core histone genes H4, H2B, H2A and H3 was amplified and identified by the techniques of homology cloning and genomic DNA walking. All the histone genes in the cluster had the structures in their 3' flanking region which related to the evolution of histone gene expression patterns throughout the cell cycle, including two different termination signals, the hairpin structure and at least one AATAAA polyadenylation signal. In their 5' region, the transcription initiation sites with a conserved sequence of 5'-PyATTCPu-3' known as the CAP site were present in all genes except to H2B, generally 37-45 bp upstream of the start code. Canonical TATA and CAAT boxes were identified only in certain histone genes. In the case of the promoters of H2B and H2A genes, there was a 5'-GATCC-3' element, which had been found to be essential to start transcription at the appropriate site. After this element, in the promoter of H2B, there was another sequence, 5'-GGATCGAAACGTTC-3', which was similar to the consensus sequence of 5'-GGAATAAACGTATTC-3' corresponding to the H2B-specific promoter element. The presence of enhancer sequences (5'-TGATATATG-3') was identified from the H4 and H3 genes, matching perfectly with the consensus sequence defined for histone genes. There were several slightly more complex repetitive DNA in the intergene regions. The presence of the series of conserved sequences and reiterated sequences was consistent with the view that mollusc histone gene cluster arose by duplicating of an ancestral precursor histone gene, the birth-and-death evolution model with strong purifying selection enabled the histone cluster less variation and more conserved function. Meanwhile, the H2A and the H2B were demonstrated to be potential good marks for phylogenetic analysis. All the results will be contributed to the characterization of repeating histone gene families in molluscs.

Real-Time Study of Genomic DNA Structural Changes upon Interaction with Small Molecules Using Dual-Polarization Interferometry

The structural changes of genomic DNA upon interaction with small molecules have been studied in real time using dual-polarization interferometry (DPI). Native or thermally denatured DNA was immobilized on the silicon oxynitride surface via a preadsorbed poly(ethylenimine) (PEI) layer. The mass loading was similar for both types of DNA, however, native DNA formed a looser and thicker layer due to its rigidity, unlike the more flexible denatured DNA, which mixed with PEI to form a denser and thinner layer. Ethidium bromide (EtBr), a classical intercalator, induced the large thickness decrease and density increase of native DNA (double-stranded), but a slight increase in both the thickness and density of denatured DNA (partial single-stranded).

Uncovering the rules for protein-protein interactions from yeast genomic data

Identifying protein-protein interactions is crucial for understanding cellular functions. Genomic data provides opportunities and challenges in identifying these interactions. We uncover the rules for predicting protein-protein interactions using a frequent pattern tree (FPT) approach modified to generate a minimum set of rules (mFPT), with rule attributes constructed from the interaction features of the yeast genomic data. The mFPT prediction accuracy is benchmarked against other commonly used methods such as Bayesian networks and logistic regressions under various statistical measures. Our study indicates that mFPT outranks other methods in predicting the protein-protein interactions for the database used. We predict a new protein-protein interaction complex whose biological function is related to premRNA splicing and new protein-protein interactions within existing complexes based on the rules generated.

The genomic structure, alternative splicing and immune response of Chlamys farreri thioester-containing protein

MEP is a member of thioester-containing protein (TEP) family found in Zhikong scallop Chlamys farreri and is involved in innate immunity against invading microbes. In the present study, the genomic DNA of CfTEP was cloned and characterized. The genomic DNA sequence of CfTEP consisted of 40 exons and 39 introns spanning 35 kb with all exon-intron junction sequences agreeing with the GT/AG consensus. The genomic organization of CfTEP was similar to human and mouse 0 rather than ciona C3-1 and Drosophila dTEP2. By RT-PCR technique, seven different cDNA variants of CfTEP (designated as CfTEP-A-CfTEP-G) were cloned from scallop gonad. CfTEP-A-CfTEP-F were produced by alternative splicing of six mutually exclusive exons (exons 19-24), respectively, which encoded the highly variable central region. While in CfTEP-G, the deletion of all the six exons introduced a new translation stop site and might trigger nonsense mediated decay (NMD). The mRNA expression and the proportion of the seven CfTEP variant transcripts were examined in the gonad of scallops after bacterial challenge. The fragments containing the highly variable central region of UTEP were amplified by RT-PCR and a 100 positive clones were sequenced randomly. The expression profiles of the seven MEP variants were different and displayed the sex and bacteria dependent manner. In the blank, sea water and Listonella anguillarum challenged subgroups of male scallops, all the transcripts detected were CfTEP-G isoform. In the Micrococcus luteus challenged subgroup, the isoforms expressed and their proportions were CfTEP-F (54%), CfTEP-B (23%), CfTEP-A (10%), CfTEP-C (7%) and CfTEP-E (6%). However, in the gonad of female scallops, only CfTEP-A were found in blank and sea water challenged subgroups. After L anguillarum or M. luteus challenge, four and five isoforms were detected, respectively, with CfTEP-F isoform being the most one in the both subgroups. These results suggested that the evolution of TEP genes was very complex, and that the diverse CfTEP transcripts generated by alternative splicing played an important role as pattern recognition receptors in the innate immune defense of scallops. (C) 2009 Elsevier Ltd. All rights reserved.

Genomic cloning, expression and recombinant protein purification of alpha and beta subunits of the allophycocyanin gene (apc) from the cyanobacterium Anacystis nidulans UTEX 625

A genomic fragment encoding alpha(APC) and beta(APC) (i.e., alpha and beta units of the allophycocyanin, APC) from Anacystis nidulans UTEX 625 was cloned and sequenced. This fragment, containing a non-coding sequence of 56 nucleotides in between, was then subcloned into the expression vector pMal-c2 downstream from and in frame with the malE gene of E. coli encoding MBP ( maltose binding protein). The fusion protein was purified by amylose affinity chromatography and cleaved by coagulation factor Xa. alpha(APC) and beta(APC) were then separated from MBP and MBP fusion proteins, respectively, and concentrated by membrane centrifugation. The study provides a method to produce recombinant allophycocyanin subunits for biomedical and biotechnological applications.

Molecular cloning, genomic organization and functional analysis of an anti-lipopolysaccharide factor from Chinese mitten crab Eriocheir sinensis

Anti-lipopolysaccharide factor (ALF) represents one kind of basic proteins, which binds and neutralizes LPS and exhibits strong antibacterial activity against Gram-negative R-type bacteria. The ALF gene of Chinese mitten crab Eriocheir sinensis (Milne Edwards, 1853) (denoted as EsALF) was identified from haemocytes by expressed sequence tag (EST) and PCR approaches. The full-length cDNA of EsALF consisted of 700 nucleotides with a canonical polyadenylation signal-sequence AATAAA, a polyA tail, and an open-reading frame of 363 bp encoding 120 amino acids. The high similarity of EsALF-deduced amino acid sequence shared with the ALFs from other species indicated that EsALF should be a member of ALF family. The mRNA expression of EsALF in the tissues of heart, gonad, gill, haemocytes, eyestalk and muscle was examined by Northern blot analysis and mRNA transcripts of EsALF were mainly detected in haemocytes, heart and gonad. The temporal expression of EsALF in haemocytes after Vibrio anguillarum challenge was recorded by quantitative real-time RT-PCR. The relative expression level of EsALF was up-regulated rapidly at 2 h post-injection and reached 3-fold to that in blank group. After a drastic decrease to the original level from 4 to 8h, the expression level increased again and reached 4-fold to that in the blank group at 12 h post-injection. The genomic DNA sequence of EsALF gene consists of 1174bp containing three exons and two introns. The coding sequence of the EsALF mature peptide was cloned and expressed in Escherichia coli BL21(DE3)-pLysS to further elucidate its biological functions. The purified recombinant product showed bactericidal activity against both Gram-positive (G(+)) and Gram-negative (G(-)) bacteria, which demonstrated that the rEsALF was a broad-spectrum antibacterial peptide. All these results indicated that EsALF was an acute-phase protein involved in the immune responses of Chinese mitten crab, and provided a potential therapeutic agent for disease control in aquaculture. (c) 2007 Elsevier Ltd. All rights reserved.

Genome constitutions of Hystrix patula, H-duthiei ssp duthiei and H-duthiei ssp longearistata (Poaceae : Triticeae) revealed by meiotic pairing behavior and genomic in-situ hybridization

Genomic constitutions of three taxa of Hystrix Moench, H. patula, H. duthiei ssp. duthiei and H. duthiei ssp. longearistata, were examined by meiotic pairing behavior and genomic in-situ hybridization. Meiotic pairing in hybrids of H. patula x Pseudoroegneria spicata (St), H. patula x Elymus wawawaiensis (StH), H. patula x H. duthiei ssp. longearistata, H. patula x Psathyrostachys huashanica (Ns(h)), H. duthiei ssp. duthiei x Psa. huashanica, H. duthiei ssp. longearistata x Psa. huashanica, Leymus multicaulis (NsXm) x H. duthiei ssp. longearistata averaged 6.53, 12.83, 1.32, 0.29, 5.18, 5.11 and 10.47 bivalents per cell, respectively. The results indicate that H. patula has the StH genome and H. duthiei ssp. duthiei and H. duthiei ssp. longearistata have the NsXm genome. Results of genomic in-situ hybridization analysis strongly supported the chromosome pairing data; therefore it is concluded that the type species of Hystrix, H. patula, should be included in Elymus, and that H. duthiei ssp. duthiei and H. duthiei ssp. longearistata should be transferred to Leymus.

Phenotype consequences of myophosphorylase dysfunction: insights from the McArdle mouse model

McArdle disease, caused by inherited deficiency of the enzyme muscle glycogen phosphorylase (GP-MM), is arguably the paradigm of exercise intolerance. The recent knock-in (p.R50X/p.R50X) mouse disease model allows an investigation of the phenotypic consequences of muscle glycogen unavailability and the physiopathology of exercise intolerance. We analysed, in 2-month-old mice [wild-type (wt/wt), heterozygous (p.R50X/wt) and p.R50X/p.R50X)], maximal endurance exercise capacity and the molecular consequences of an absence of GP-MM in the main glycogen metabolism regulatory enzymes: glycogen synthase, glycogen branching enzyme and glycogen debranching enzyme, as well as glycogen content in slow-twitch (soleus), intermediate (gastrocnemius) and glycolytic/fast-twitch (extensor digitorum longus; EDL) muscles.

Integration of genomic and phenotypic data

Clare, A. (2005) Integration of genomic and phenotypic data. In Data Analysis and Visualization in Genomics and Proteomics, Eds. Francisco Azuaje and Joaquin Dopazo, Wiley, London. ISBN: 0-470-09439-7

Functional genomic hypothesis generation and experimentation by a robot scientist

King R. D., Whelan, K. E., Jones, F. M., Reiser, P. G. K., Bryant, C. H., Muggleton, S., Kell, D. B. and Oliver, S. G. (2004) Functional genomic hypothesis generation and experimentation by a robot scientist. Nature 427 (6971) p247-252

Machine learning of functional class from phenotype data

Clare, A. and King R.D. (2002) Machine learning of functional class from phenotype data. Bioinformatics 18(1) 160-166