Michael Molla

Genome-wide in situ exon capture for selective resequencing

Increasingly powerful sequencing technologies are ushering in an era of personal genome sequences and raising the possibility of using such information to guide medical decisions. Genome resequencing also promises to accelerate the identification of disease-associated mutations. Roughly 98% of the human genome is composed of repeats and intergenic or non–protein-coding sequences. Thus, it is crucial to focus resequencing on high-value genomic regions. Protein-coding exons represent one such type of high-value target. We have developed a method of using flexible, high-density microarrays to capture any desired fraction of the human genome, in this case corresponding to more than 200,000 protein-coding exons. Depending on the precise protocol, up to 55–85% of the captured fragments are associated with targeted regions and up to 98% of intended exons can be recovered. This methodology provides an adaptable route toward rapid and efficient resequencing of any sizeable, non-repeat portion of the human genome.

Complete genome sequence of Treponema pallidum ssp. pallidum strain SS14 determined with oligonucleotide arrays

BackgroundSyphilis spirochete Treponema pallidum ssp. pallidum remains the enigmatic pathogen, since no virulence factors have been identified and the pathogenesis of the disease is poorly understood. Increasing rates of new syphilis cases per year have been observed recently.ResultsThe genome of the SS14 strain was sequenced to high accuracy by an oligonucleotide array strategy requiring hybridization to only three arrays (Comparative Genome Sequencing, CGS). Gaps in the resulting sequence were filled with targeted dideoxy-terminators (DDT) sequencing and the sequence was confirmed by whole genome fingerprinting (WGF). When compared to the Nichols strain, 327 single nucleotide substitutions (224 transitions, 103 transversions), 14 deletions, and 18 insertions were found. On the proteome level, the highest frequency of amino acid-altering substitution polymorphisms was in novel genes, while the lowest was in housekeeping genes, as expected by their evolutionary conservation. Evidence was also found for hypervariable regions and multiple regions showing intrastrain heterogeneity in the T. pallidum chromosome.ConclusionThe observed genetic changes do not have influence on the ability of Treponema pallidum to cause syphilitic infection, since both SS14 and Nichols are virulent in rabbit. However, this is the first assessment of the degree of variation between the two syphilis pathogens and paves the way for phylogenetic studies of this fascinating organism.

Using machine learning to design and interpret gene-expression microarrays

Gene-expression microarrays, commonly called gene chips, make it possible to simultaneously measure the rate at which a cell or tissue is expressing--translating into a protein--each of its thousands of genes. One can use these comprehensive snapshots of biological activity to infer regulatory pathways in cells; identify novel targets for drug design; and improve the diagnosis, prognosis, and treatment planning for those suffering from disease. However, the amount of data this new technology produces is more than one can manually analyze. Hence, the need for automated analysis of microarray data offers an opportunity for machine learning to have a significant impact on biology and medicine. This article describes microarray technology, the data it produces, and the types of machine learning tasks that naturally arise with these data. It also reviews some of the recent prominent applications of machine learning to gene-chip data, points to related tasks where machine learn-Lug might have a further impact on biology and medicine, and describes additional types of interesting data that recent advances in biotechnology allow biomedical researchers to collect.

COMPENSATORY EVOLUTION OF INTERACTING GENE PRODUCTS THROUGH MULTIFUNCTIONAL INTERMEDIATES

Abstract When two mutations are singly deleterious but neutral or beneficial together, compensatory evolution can occur. The accumulation of derived, compensated genotypes contributes to the evolution of genetic incompatibilities between diverging populations or species. Previous two locus/two allele models have shown that compensatory evolution is appreciable only with tight linkage, the possibility of nearly simultaneous mutations, and/or a way to overcome negative selection against the singly mutated genotype. These conditions are often not met. Even when they are met, compensatory evolution is still predicted to be extremely slow, and in many scenarios selective advantage of the compensated genotype does little to accelerate it. Despite these obstacles, empirical studies suggest that it occurs readily. We describe here a set of related two locus/three allele models that invoke plausible neutral intermediates capable of productive interaction with both ancestral and compensated products of the interacting locus. These models are explored with analytical and computer simulation methods. The effect of these stepping‐stone alleles on the evolution of ancestor‐descendant incompatibilities is often profound, making the difference between evolution and stasis in several situations, including in small populations, when codominance or haploidy prevents shielding of mismatched genotypes, and in the absence of positive selection on the derived genotype. However, in large populations these intermediates can either speed or slow the evolution of incompatible genotypes relative to the two‐allele case, depending on the specific fitness model. These results suggest that population size, the source of adaptive benefit, and the structural details of heteromeric gene product complexes interact to influence the path by which intergenic incompatibility evolves.

Interpreting microarray expression data using text annotating the genes

Microarray expression data is being generated by the gigabyte all over the world with undoubted exponential increases to come. Annotated genomic data is also rapidly pouring into public databases. Our goal is to develop automated ways of combining these two sources of information to produce insight into the operation of cells under various conditions. Our approach is to use machine-learning techniques to identify characteristics of genes that are up-regulated or down-regulated in a particular micro-array experiment. We seek models that are (a) accurate. (b) easy to interpret, and (c) stable to small variations in the training data. This paper explores the effectiveness of two standard machine-learning algorithms for this task: Naive Bayes (based on probability) and PFOIL (based on building rules). Although we do not anticipate using our learned models to predict expression levels of genes, we cast the task in a predictive framework, and evaluate the quality of the models in terms of their predictive power on genes held out from the training. The paper reports on experiments using actual E. coli microarray data, discussing the strengths and weaknesses of the two algorithms and demonstrating the trade-offs between accuracy, comprehensibility, and stability.