PubMed Health⌕ Search

Biomedical subjects

Shea N Gardner

Publications and source records attributed to Shea N Gardner.

11 recordsLinked to original sources
PubMed2005

Draft versus finished sequence data for DNA and protein diagnostic signature development.

Sequencing pathogen genomes is costly, demanding careful allocation of limited sequencing resources. We built a computational Sequencing Analysis Pipeline (SAP) to guide decisions regarding the amount of genomic sequencing necessary to develop high-quality diagnostic DNA and protein signatures. SAP uses simulations to estimate the number of target genomes and close phylogenetic relatives (near neighbors or NNs) to sequence. We use SAP to assess whether draft data are sufficient or finished sequencing is required using Marburg and variola virus sequences. Simulations indicate that intermediate to high-quality draft with error rates of 10(-3)-10(-5) (approximately 8x coverage) of target organisms is suitable for DNA signature prediction. Low-quality draft with error rates of approximately 1% (3x to 6x coverage) of target isolates is inadequate for DNA signature prediction, although low-quality draft of NNs is sufficient, as long as the target genomes are of high quality. For protein signature prediction, sequencing errors in target genomes substantially reduce the detection of amino acid sequence conservation, even if the draft is of high quality. In summary, high-quality draft of target and low-quality draft of NNs appears to be a cost-effective investment for DNA signature prediction, but may lead to underestimation of predicted protein signatures.

Computational Biology↗
PubMed2005

Software for optimization of SNP and PCR-RFLP genotyping to discriminate many genomes with the fewest assays.

BACKGROUND: Microbial forensics is important in tracking the source of a pathogen, whether the disease is a naturally occurring outbreak or part of a criminal investigation. RESULTS: A method and SPR Opt (SNP and PCR-RFLP Optimization) software to perform a comprehensive, whole-genome analysis to forensically discriminate multiple sequences is presented. Tools for the optimization of forensic typing using Single Nucleotide Polymorphism (SNP) and PCR-Restriction Fragment Length Polymorphism (PCR-RFLP) analyses across multiple isolate sequences of a species are described. The PCR-RFLP analysis includes prediction and selection of optimal primers and restriction enzymes to enable maximum isolate discrimination based on sequence information. SPR Opt calculates all SNP or PCR-RFLP variations present in the sequences, groups them into haplotypes according to their co-segregation across those sequences, and performs combinatoric analyses to determine which sets of haplotypes provide maximal discrimination among all the input sequences. Those set combinations requiring that membership in the fewest haplotypes be queried (i.e. the fewest assays be performed) are found. These analyses highlight variable regions based on existing sequence data. These markers may be heterogeneous among unsequenced isolates as well, and thus may be useful for characterizing the relationships among unsequenced as well as sequenced isolates. The predictions are multi-locus. Analyses of mumps and SARS viruses are summarized. Phylogenetic trees created based on SNPs, PCR-RFLPs, and full genomes are compared for SARS virus, illustrating that purported phylogenies based only on SNP or PCR-RFLP variations do not match those based on multiple sequence alignment of the full genomes. CONCLUSION: This is the first software to optimize the selection of forensic markers to maximize information gained from the fewest assays, accepting whole or partial genome sequence data as input. As more sequence data becomes available for multiple strains and isolates of a species, automated, computational approaches such as those described here will be essential to make sense of large amounts of information, and to guide and optimize efforts in the laboratory. The software and source code for SPR Opt is publicly available and free for non-profit use at http://www.llnl.gov/IPandC/technology/software/softwaretitles/spropt.php.

Cluster Analysis↗
PubMed2005

System to assess genome sequencing needs for viral protein diagnostics and therapeutics.

Computational analyses of genome sequences may elucidate protein signatures unique to a target pathogen. We constructed a Protein Signature Pipeline to guide the selection of short peptide sequences to serve as targets for detection and therapeutics. In silico identification of good target peptides that are conserved among strains and unique compared to other species generates a list of peptides. These peptides may be developed in the laboratory as targets of antibody, peptide, and ligand binding for detection assays and therapeutics or as targets for vaccine development. In this paper, we assess how the amount of sequence data affects our ability to identify conserved, unique protein signature candidates. To determine the amount of sequence data required to select good protein signature candidates, we have built a computationally intensive system called the Sequencing Analysis Pipeline (SAP). The SAP performs thousands of Monte Carlo simulations, each calling the Protein Signature Pipeline, to assess how the amount of sequence data for a target organism affects the ability to predict peptide signature candidates. Viral species differ substantially in the number of genomes required to predict protein signature targets. Patterns do not appear based on genome structure. There are more protein than DNA signatures due to greater intraspecific conservation at the protein than at the nucleotide level. We conclude that it is necessary to use the SAP as a dynamic system to assess the need for continued sequencing for each species individually and to update predictions with each additional genome that is sequenced.

Base Sequence↗
PubMed2004

Sequencing needs for viral diagnostics.

We built a system to guide decisions regarding the amount of genomic sequencing required to develop diagnostic DNA signatures, which are short sequences that are sufficient to uniquely identify a viral species. We used our existing DNA diagnostic signature prediction pipeline, which selects regions of a target species genome that are conserved among strains of the target (for reliability, to prevent false negatives) and unique relative to other species (for specificity, to avoid false positives). We performed simulations, based on existing sequence data, to assess the number of genome sequences of a target species and of close phylogenetic relatives (near neighbors) that are required to predict diagnostic signature regions that are conserved among strains of the target species and unique relative to other bacterial and viral species. For DNA viruses such as variola (smallpox), three target genomes provide sufficient guidance for selecting species-wide signatures. Three near-neighbor genomes are critical for species specificity. In contrast, most RNA viruses require four target genomes and no near-neighbor genomes, since lack of conservation among strains is more limiting than uniqueness. Severe acute respiratory syndrome and Ebola Zaire are exceptional, as additional target genomes currently do not improve predictions, but near-neighbor sequences are urgently needed. Our results also indicate that double-stranded DNA viruses are more conserved among strains than are RNA viruses, since in most cases there was at least one conserved signature candidate for the DNA viruses and zero conserved signature candidates for the RNA viruses.

Base Sequence↗
PubMed2004

Cytostatic anticancer drug development.

This review focuses on clinical trials and the approval process in order to understand the discrepancy between vibrant science and the continuing failure of mechanism-based anticancer drugs. CLINICAL TRIALS: Mechanistic trials in cancer require at least three elements: the assurance of tumor definition, knowledge of the natural history, and earlier intervention. Histologic identity is not a reliable surrogate of the functional nature or a predictor of the natural history. cDNA arrays and computational models have promise in improving diagnosis and prediction, and thereby making tailored therapy possible. The latter requires: the incorporation of initial and earlier rational combination therapy, dynamic models of disease progression, and methods to discourage the emergence of resistance. For cytostatics, and in early cancer, a delay in progression may represent a better index of survival than tumor shrinkage. APPROVAL PROCESS: Since mechanistic similarities may outweigh therapeutic predictions based on organ-and histology-defined cancer, there is a need for a revised process that would allow for tailored treatment and initial combination therapy to improve safety, efficacy, and circumvent resistance. CONCLUSION: In order to translate the major and immediate potential of cytostatic drugs, clinical trials and the approval process may need to shift to a mechanism-based framework.

Antineoplastic Agents↗
PubMed2003

Limitations of TaqMan PCR for detecting divergent viral pathogens illustrated by hepatitis A, B, C, and E viruses and human immunodeficiency virus.

Recent events illustrate the imperative to rapidly and accurately detect and identify pathogens during disease outbreaks, whether they are natural or engineered. Particularly for our primary goal of detecting bioterrorist releases, detection techniques must be both species-wide (capable of detecting all known strains of a given species) and species specific. Due to classification restrictions on the publication of data for species that may pose a bioterror threat, we illustrate the challenges of finding such assays using five nonthreat organisms that are nevertheless of public health concern: human immunodeficiency virus (HIV) and four species of hepatitis viruses. Fluorogenic probe-based PCR assays (TaqMan; Perkin-Elmer Corp., Applied Biosystems, Foster City, Calif.) may be sensitive, fast methods for the identification of species in which the genome is conserved among strains, such as hepatitis A virus. For species such as HIV, however, the strains are highly divergent. We use computational methods to show that nine TaqMan primer and probe sequences, or signatures, are needed to ensure that all strains will be detected, but this is an unfeasible number, considering the cost of TaqMan probes. Strains of hepatitis B, C, and E viruses show intermediate divergence, so that two to three TaqMan signatures are required to detect all strains of each virus. We conclude that for species such as hepatitis A virus with high levels of sequence conservation among strains, signatures can be found computationally for detection by the TaqMan assay, which is a sensitive, rapid, and cost-effective method. However, for species such as HIV with substantial genetic divergence among strains, the TaqMan assay becomes unfeasible and alternative detection methods may be required. We compare the TaqMan assay with some of the alternative nucleic acid-based detection techniques of microarray, chip, and bead technologies in terms of sensitivity, speed, and cost.

Computational Biology↗
PubMed2003

New tools for cancer chemotherapy: computational assistance for tailoring treatments.

Computational models of cancer chemotherapy have the potential to streamline clinical trial design, contribute to the design of rational, tailored treatments, and facilitate our understanding of experimental results. Mechanistic models based on functional data from tumor biopsies will enable physicians to predict response to treatment for a specific patient, in contrast to statistical models in which the probability of response for a given patient may differ substantially from the population average. While microarray analyses of gene expression also show promise for guiding individualized treatments, it may be difficult to link statistical mining of microarray data with mechanistic, tailored treatments. Furthermore, gene expression does not identify how drugs should be scheduled. This review summarizes mechanistic mathematical models developed to improve the design of chemotherapy regimens. Mechanistic models that incorporate both genetic resistance and cell cycle-mediated resistance during treatment with multiple drugs will be most useful in designing treatment regimens tailored for individuals. Because there are already a number of papers that address the applications of microarray technology, we will limit our discussion to the contrasts between mechanistic computational models and microarray technology, and how these two approaches may complement one another.

Drug Resistance, Neoplasm↗
PubMed2002

Modeling multi-drug chemotherapy: tailoring treatment to individuals.

BACKGROUND: Predicting and tailoring optimal cancer treatments presents a major challenge. METHODS: A computational model (kinetically tailored treatment, or KITT model) is developed to predict drug combinations, doses, and schedules likely to be effective in reducing tumor size and prolonging patient life. Treatment strategies may be tailored to individuals based on tumor cell kinetics. The model incorporates intra-tumor heterogeneity and evolution of drug resistance, apoptotic rates, and cell division rates. Tumor growth may follow an exponential or a Gompertzian trajectory. Drug pharmacodynamic and pharmacokinetic models are used. Toxicity is modeled in several ways. RESULTS: A key prediction of KITT is that including cytostatic drugs like tamoxifen and herceptin during treatment with cytotoxic drugs substantially increases the probability of cure and prolongs patient life. Results also suggest that altering drug scheduling may be more effective but not more toxic than dose escalation. CAF chemotherapy (cyclophosphamide, adriamycin, and 5-fluorouracil) is predicted to be more effective than CMF (cyclophosphamide, methotrexate, and 5-fluorouracil). KITT also suggests that tumors with a high proliferative index (PI) may respond better to drug combinations incorporating two cell-cycle phase-specific drugs than do tumors with a low PI. Tumors with a low PI, in contrast, are predicted to respond better to regimens involving two cell-cycle phase-non-specific drugs than do tumors with a high PI. These predictions are borne out by clinical trial results published in the literature, which are discussed. Simulated predictions of the model match well with results from a clinical trial by Silvestrini et al. (2000. Int. J. Cancer 87, 405). The results of simulating the growth of 26896 tumors are used to construct a decision tree for prognosis to identify the key tumor and treatment variables. CONCLUSION: Additional tests of the model are needed in which physicians collect information on apoptotic and proliferative indices, cell-cycle times, and drug resistance from biopsies of each individual's tumor. Computational models may become important tools to help optimize and tailor cancer treatments.

Antineoplastic Combined Chemotherapy Protocols↗
PubMed2002

The cellular lethality of radiation-induced chromosome translocations in human lymphocytes.

Recent evidence has shown that translocation frequencies decline over time. This phenomenon might be explained by the co-occurrence of translocations in cells that also contain dicentrics, in which case translocations would be eliminated as a by-product of selection against dicentrics. Alternatively, a fraction of translocations may themselves be lethal. Here we describe our initial approaches to develop mathematical models to test whether the decline in translocation frequencies results from the first, the second, or a combination of these two possibilities. The models assumed that all chromosome exchanges were simple, i.e., were comprised of dicentrics as well as one-way and two-way translocations. Complex aberrations (three or more breaks in two or more chromosomes) were not modeled, nor were fragments or intrachromosomal exchanges (rings, inversions). We tested the models using Monte Carlo simulations, and then we fitted the models to data describing chromosome aberration frequencies induced by a single acute in vitro exposure to (137)Cs gamma rays in human peripheral blood lymphocytes from two donors. Chromosome painting was used to enumerate translocations and dicentrics from 2 to 7 days after exposure. Our results indicate that in donor no. 2, the decline in translocation frequencies occurs as a by-product of selection against dicentrics. However, in donor no. 1, whose cells appeared more radiosensitive than cells from donor no. 2, up to 40% of the one-way translocations may themselves be lethal at high doses, although calculations indicate that two-way translocations do not cause lymphocyte mortality. Individual variation in the probability that translocations are lethal to cells appears to be important, and one-way translocations appear to be lethal more often than two-way translocations. Within the limits of these models, these findings indicate that both postulated mechanisms, i.e. inherent lethality and selection against dicentrics in the same cells, contribute to the loss of both one-way and two-way translocations.

Cell Death↗
PubMedDate not supplied

Cell cycle phase-specific chemotherapy: computational methods for guiding treatment.

Computational models of cancer chemotherapy enhance the understanding of in vitro, in vivo, and clinical trial data and have the potential to contribute to the design of rational treatment regimens. In particular, mechanistic, predictive models are superior to statistical, phenomenological descriptions of data. Mechanistic models based on functional data from tumor biopsies will enable the response to treatment to be predicted for a specific patient, in contrast to statistical models in which the probability of response for a given patient may differ substantially from the population average. This review summarizes mathematical models developed to improve the design of treatment regimens using cell-cycle phase-specific chemotherapy. It starts with simple models of dose response, then moves to more complex models of scheduling cell-cycle phase-specific drugs, and finally discusses mechanistic models that incorporate both genetic drug resistance and cell cycle-mediated drug resistance. This last class of models will be most useful in designing treatment regimens tailored for individual patients.

Animals↗