PowerPoint Presentation

The PinesOctober 28, 2013Genomic MedicineMalcolm CampbellJames G. Martin Genomics Program DirectorProfessor of Biology, Davidson College

Outline of TalkWhat is a genome?How to sequence genomes?Diagnose and Treat Cancers Better?New Drugs from Failures: IressaNew Treatment ParadigmsGut microbiomeAdenineThymineGuanine Cytosine2Science PresentationGive you the data, help you interpret.AdenineThymineGuanine Cytosine3What is a Genome?

AdenineThymineGuanine Cytosine4Genetic information that you uniqueHuman genome 3.4 billion base pairsG:C or A:T base pairs~23,000 genes

What is a Genome?AdenineThymineGuanine Cytosine5If the human genome were compiled in books:200 volumes, 1000 pages eachread 10 bases/second = 315,360,000 bases/year9.5 years to read out loud (without stopping)

What is the Human Genome?AdenineThymineGuanine Cytosine6Determine order of bases on all 23 (24) chromosomesCan only read 30 to 700 bases at a timeCannot sequence a genome in one runWhole Genome Shotgun sequencingHow do you sequence a genome?Not like book, can only read 30 to 700 bases @ time7Determine order of bases on all 23 (24) chromosomesCan only read 30 to 700 bases at a timeCannot sequence a genome in one runWhole Genome Shotgun sequencingHow do you sequence a genome?Not like book, can only read 30 to 700 bases @ time8

Modern Genome SequencingNot like book, can only read 30 to 700 bases @ time9This is an analogy for genome assembly. This page will be torn into horizontal strips.Modern Genome SequencingNot like book, can only read 30 to 700 bases @ time10Modern Genome Sequencingis ans anThis iNot like book, can only read 30 to 700 bases @ time11Modern Genome Sequencingis as anThis iNot like book, can only read 30 to 700 bases @ time12Modern Genome Sequencingis as anThis iThis is an analogy for genome assembly. This page will be torn into horizontal strips.Not like book, can only read 30 to 700 bases @ time13You dont know the language or syntax.is as anThis iNot like book, can only read 30 to 700 bases @ time14Huge Pile of Strips

is as anThis iNot like book, can only read 30 to 700 bases @ time15Needle in a Haystack?

is as anThis iNot like book, can only read 30 to 700 bases @ time16Reassemble the Tree from Paper

Not like book, can only read 30 to 700 bases @ time17Modern Genome Sequence Assemblymultiple copies of genomeconsensus genome sequencealigned sequencing readsNot like book, can only read 30 to 700 bases @ time18Modern Genome Sequence Assemblymultiple copies of genomeconsensus genome sequencealigned sequencing readsNot like book, can only read 30 to 700 bases @ time19Modern Genome Sequence Assemblymultiple copies of genomeconsensus genome sequencealigned sequencing readshigh coveragelow coverageNot like book, can only read 30 to 700 bases @ time20Modern Genome Sequence Assemblymultiple copies of genomeconsensus genome sequencealigned sequencing readsNot like book, can only read 30 to 700 bases @ time21Reference Genome Assembly ATGGCATTGCAA TGGCATTGCAATTTGAGATGGTATTG GATGGCATTGCAA GCATTGCAATTTGAC ATGGCATTGCAATTTAGATGGTATTGCAATTTGNot like book, can only read 30 to 700 bases @ time22Reference Genome Assembly ATGGCATTGCAA TGGCATTGCAATTTGAGATGGTATTG GATGGCATTGCAA GCATTGCAATTTGAC ATGGCATTGCAATTTAGATGGTATTGCAATTTGconsensus AGATGGCATTGCAATTTGACNot like book, can only read 30 to 700 bases @ time23

206 bones> 60 organsHow do I remember them all?Not like book, can only read 30 to 700 bases @ time24

206 bones> 60 organsThink like a genomicist

Not like book, can only read 30 to 700 bases @ time251 22 + X & Y

AdenineThymineGuanine Cytosine26Even EasierGCATAdenineThymineGuanine Cytosine27Is there a better way to diagnose and treat cancers?

28Figure 7.1 DLBCL gene expression analysis.a) Dendrogram shows the gene expression relatedness of the 96samples used in this study. Labels A and G refer to the samples that express activated B-cell signature genes and germinal center B-cell signatures, respectively. The color key for cell types is the same for both halves of the figure. b) Hierarchical clustering shows the 17,856rows of genes and 96 columns of samples tested. Each colored rectangle in the clustering represents an expression ratio, with red indicating induction in the samples, and green repression. The numbers indicate relative expression on a log2 scale on top (-2 to +2)and fluorescence ratios below (0.254.0). Experimental samples were used to produce the red (Cy5) cDNAs, which were compared to a mixture of green (Cy3) cDNAs produced from a pool of nine different lymphoma cell lines.Is there a better way to diagnose and treat cancers?

Diffuse Large B Cell Lymphoma(DLBCL)29Figure 7.1 DLBCL gene expression analysis.a) Dendrogram shows the gene expression relatedness of the 96samples used in this study. Labels A and G refer to the samples that express activated B-cell signature genes and germinal center B-cell signatures, respectively. The color key for cell types is the same for both halves of the figure. b) Hierarchical clustering shows the 17,856rows of genes and 96 columns of samples tested. Each colored rectangle in the clustering represents an expression ratio, with red indicating induction in the samples, and green repression. The numbers indicate relative expression on a log2 scale on top (-2 to +2)and fluorescence ratios below (0.254.0). Experimental samples were used to produce the red (Cy5) cDNAs, which were compared to a mixture of green (Cy3) cDNAs produced from a pool of nine different lymphoma cell lines.Diffuse Large B Cell Lymphoma(DLBCL)25,000 new cases each year in AmericaHalf of patients die despite chemotherapyWhy subject them to chemo if not helpful?30Figure 7.1 DLBCL gene expression analysis.a) Dendrogram shows the gene expression relatedness of the 96samples used in this study. Labels A and G refer to the samples that express activated B-cell signature genes and germinal center B-cell signatures, respectively. The color key for cell types is the same for both halves of the figure. b) Hierarchical clustering shows the 17,856rows of genes and 96 columns of samples tested. Each colored rectangle in the clustering represents an expression ratio, with red indicating induction in the samples, and green repression. The numbers indicate relative expression on a log2 scale on top (-2 to +2)and fluorescence ratios below (0.254.0). Experimental samples were used to produce the red (Cy5) cDNAs, which were compared to a mixture of green (Cy3) cDNAs produced from a pool of nine different lymphoma cell lines.Who will benefit from chemotherapy?Brown and Botstein31Figure 7.1 DLBCL gene expression analysis.a) Dendrogram shows the gene expression relatedness of the 96samples used in this study. Labels A and G refer to the samples that express activated B-cell signature genes and germinal center B-cell signatures, respectively. The color key for cell types is the same for both halves of the figure. b) Hierarchical clustering shows the 17,856rows of genes and 96 columns of samples tested. Each colored rectangle in the clustering represents an expression ratio, with red indicating induction in the samples, and green repression. The numbers indicate relative expression on a log2 scale on top (-2 to +2)and fluorescence ratios below (0.254.0). Experimental samples were used to produce the red (Cy5) cDNAs, which were compared to a mixture of green (Cy3) cDNAs produced from a pool of nine different lymphoma cell lines.

Who will benefit from chemotherapy?control setpatient and control biopsies32Figure 7.1 DLBCL gene expression analysis.a) Dendrogram shows the gene expression relatedness of the 96samples used in this study. Labels A and G refer to the samples that express activated B-cell signature genes and germinal center B-cell signatures, respectively. The color key for cell types is the same for both halves of the figure. b) Hierarchical clustering shows the 17,856rows of genes and 96 columns of samples tested. Each colored rectangle in the clustering represents an expression ratio, with red indicating induction in the samples, and green repression. The numbers indicate relative expression on a log2 scale on top (-2 to +2)and fluorescence ratios below (0.254.0). Experimental samples were used to produce the red (Cy5) cDNAs, which were compared to a mixture of green (Cy3) cDNAs produced from a pool of nine different lymphoma cell lines.

Measure Gene Activity in All Cells33Figure 7.2 Biologically distinct DLBCL gene expression signatures.Expanded view (from Figure 7.1) of biologically distinct gene expression signatures defined by hierarchical clustering. Most genes without designation on the right are genes with unknown functions.

Characterize Cancers by Gene Activity34Figure 7.3 Discovery of DLBCL subtypes.a) DLBCL samples and two normal germinal cells were clustered using the GC signature genes from Figure 7.2, forming two subtypes: GC-like DLBCLs (orange) and activated B-cell-like DLBCLs (blue). The two normal GC B-cells (black) clustered among the GC-like DLBCLs. b) The DLBCL dendrogram was maintained and used to recluster all 17,856genes to expand the 2 signature sets of genes indicated by orange and blue bars. c) The larger signature gene sets were reclustered to produce a clear distinction in the opposing profiles. The dendrogram of DLBCLs is enlarged to reveal individual sample names and branching structures.

Retrospective Clinical Outcomesbased on gene activity35Figure 7.4 Clinical distinctions of DLBCL.Kaplan-Meier plot of 40 patients being treated for DLBCL. The graph moves down a step each time a patient dies. The tick marks indicate when a patient was surveyed by a clinician. a) When the patients were separated into GC-like or activated B-cell-like categories, the survival rates were noticeably different, as indicated. b) When more traditional IPI indices were used, the patients again segregated into high-risk and low-risk categories, as indicated. The X-axis is the number of years survival postdiagnosis, and the Y-axis is the fraction of surviving patients.

Retrospective Clinical Outcomesbased on gene activitytraditional prediction36Figure 7.4 Clinical distinctions of DLBCL.Kaplan-Meier plot of 40 patients being treated for DLBCL. The graph moves down a step each time a patient dies. The tick marks indicate when a patient was surveyed by a clinician. a) When the patients were separated into GC-like or activated B-cell-like categories, the survival rates were noticeably different, as indicated. b) When more traditional IPI indices were used, the patients again segregated into high-risk and low-risk categories, as indicated. The X-axis is the number of years survival postdiagnosis, and the Y-axis is the fraction of surviving patients.

Retrospective Clinical Outcomeswrong 27%wrong 32%37Figure 7.4 Clinical distinctions of DLBCL.Kaplan-Meier plot of 40 patients being treated for DLBCL. The graph moves down a step each time a patient dies. The tick marks indicate when a patient was surveyed by a clinician. a) When the patients were separated into GC-like or activated B-cell-like categories, the survival rates were noticeably different, as indicated. b) When more traditional IPI indices were used, the patients again segregated into high-risk and low-risk categories, as indicated. The X-axis is the number of years survival postdiagnosis, and the Y-axis is the fraction of surviving patients.

Retrospective Clinical Outcomeswrong 37%38Figure 7.4 Clinical distinctions of DLBCL.Kaplan-Meier plot of 40 patients being treated for DLBCL. The graph moves down a step each time a patient dies. The tick marks indicate when a patient was surveyed by a clinician. a) When the patients were separated into GC-like or activated B-cell-like categories, the survival rates were noticeably different, as indicated. b) When more traditional IPI indices were used, the patients again segregated into high-risk and low-risk categories, as indicated. The X-axis is the number of years survival postdiagnosis, and the Y-axis is the fraction of surviving patients.

Resort Low Risk Patientswrong 37%wrong 29%39Figure 7.5 Clinical distinction of IPI low-risk DLBCL patients.The graph moves down a step each time a patient dies. The tick marks indicate when a patient was surveyed by a clinician. The 24 low-risk patients from Figure 7.4b were reanalyzed using the GC signature gene analysis and segregated into two categories as indicated.Six Key Indicator GenesGenes OnGenes OnLMO2BCL6FN1BCL2SCYA3CCND2chemono chemoCan Breast Cancer Treatment Be Improved?41Figure 7.6 Variations in expression of 1,753 human ORFs.a) Magnified view of the 65 biopsies and 17 cell lines as they were clustered based on expression profiles of 1,753 genes. Before (BE) and after (AF) tissue pairs from the same patient that clustered together are colored red; the two tumor/lymph node pairs are light blue; three normal breast samples are green; and unpaired samples are black. b) A display of every gene measured.c)j) Each row in the gene expression profiles represents a gene and each column represents a tissue sample. After the genes were clustered, the samples were clustered to arrange them according to similar expression profiles. Eight groups of 1,753 genes were identified as cell-type signatures: c) endothelial cells; d) stromal/fibroblast; e) breast basal epithelial; f) B-cell; g) adipose-enriched/normal breast; h) macrophage; I) T-cell; j) luminal epithelial.The 17 cell lines were clustered separately from tissue samples.

Can Breast Cancer Treatment Be Improved?tissue biopsies42Figure 7.6 Variations in expression of 1,753 human ORFs.a) Magnified view of the 65 biopsies and 17 cell lines as they were clustered based on expression profiles of 1,753 genes. Before (BE) and after (AF) tissue pairs from the same patient that clustered together are colored red; the two tumor/lymph node pairs are light blue; three normal breast samples are green; and unpaired samples are black. b) A display of every gene measured.c)j) Each row in the gene expression profiles represents a gene and each column represents a tissue sample. After the genes were clustered, the samples were clustered to arrange them according to similar expression profiles. Eight groups of 1,753 genes were identified as cell-type signatures: c) endothelial cells; d) stromal/fibroblast; e) breast basal epithelial; f) B-cell; g) adipose-enriched/normal breast; h) macrophage; I) T-cell; j) luminal epithelial.The 17 cell lines were clustered separately from tissue samples.

Can Breast Cancer Treatment Be Improved?43Figure 7.6 Variations in expression of 1,753 human ORFs.a) Magnified view of the 65 biopsies and 17 cell lines as they were clustered based on expression profiles of 1,753 genes. Before (BE) and after (AF) tissue pairs from the same patient that clustered together are colored red; the two tumor/lymph node pairs are light blue; three normal breast samples are green; and unpaired samples are black. b) A display of every gene measured.c)j) Each row in the gene expression profiles represents a gene and each column represents a tissue sample. After the genes were clustered, the samples were clustered to arrange them according to similar expression profiles. Eight groups of 1,753 genes were identified as cell-type signatures: c) endothelial cells; d) stromal/fibroblast; e) breast basal epithelial; f) B-cell; g) adipose-enriched/normal breast; h) macrophage; I) T-cell; j) luminal epithelial.The 17 cell lines were clustered separately from tissue samples.

No Patterns Emergedcontrol setpatient biopsies44Figure 7.6 Variations in expression of 1,753 human ORFs.a) Magnified view of the 65 biopsies and 17 cell lines as they were clustered based on expression profiles of 1,753 genes. Before (BE) and after (AF) tissue pairs from the same patient that clustered together are colored red; the two tumor/lymph node pairs are light blue; three normal breast samples are green; and unpaired samples are black. b) A display of every gene measured.c)j) Each row in the gene expression profiles represents a gene and each column represents a tissue sample. After the genes were clustered, the samples were clustered to arrange them according to similar expression profiles. Eight groups of 1,753 genes were identified as cell-type signatures: c) endothelial cells; d) stromal/fibroblast; e) breast basal epithelial; f) B-cell; g) adipose-enriched/normal breast; h) macrophage; I) T-cell; j) luminal epithelial.The 17 cell lines were clustered separately from tissue samples.

People Are More Unique Than Their Cancersbefore (BE) and after (AF) chemotherapy45Figure 7.6 Variations in expression of 1,753 human ORFs.a) Magnified view of the 65 biopsies and 17 cell lines as they were clustered based on expression profiles of 1,753 genes. Before (BE) and after (AF) tissue pairs from the same patient that clustered together are colored red; the two tumor/lymph node pairs are light blue; three normal breast samples are green; and unpaired samples are black. b) A display of every gene measured.c)j) Each row in the gene expression profiles represents a gene and each column represents a tissue sample. After the genes were clustered, the samples were clustered to arrange them according to similar expression profiles. Eight groups of 1,753 genes were identified as cell-type signatures: c) endothelial cells; d) stromal/fibroblast; e) breast basal epithelial; f) B-cell; g) adipose-enriched/normal breast; h) macrophage; I) T-cell; j) luminal epithelial.The 17 cell lines were clustered separately from tissue samples.

Characterize Cancers by Personal Differences46Figure 7.7 Cluster analysis using the intrinsic gene subset.The 496 genes were highly coordinated in each pair of before-and-after chemotherapy samples, but differed in the rest of the samples. When these genes were used to cluster the samples, a) two major branches were produced. Seventeen sample pairs were placed next to each other (small black brackets), but three were not (larger green brackets). b)f) Based on the four clusters of genes in the expression profiles, the samples were clustered into four main types: basal cell-like (orange); Erb-B2 positive (pink); normal breast tissue-like (green); luminal epithelial and estrogen receptor positive (blue). Three of the samples did not fit into any of these four categories.

Characterize Cancers by Personal Differences47Figure 7.7 Cluster analysis using the intrinsic gene subset.The 496 genes were highly coordinated in each pair of before-and-after chemotherapy samples, but differed in the rest of the samples. When these genes were used to cluster the samples, a) two major branches were produced. Seventeen sample pairs were placed next to each other (small black brackets), but three were not (larger green brackets). b)f) Based on the four clusters of genes in the expression profiles, the samples were clustered into four main types: basal cell-like (orange); Erb-B2 positive (pink); normal breast tissue-like (green); luminal epithelial and estrogen receptor positive (blue). Three of the samples did not fit into any of these four categories.

Breast Cancer is a Misnomer48Figure 7.7 Cluster analysis using the intrinsic gene subset.The 496 genes were highly coordinated in each pair of before-and-after chemotherapy samples, but differed in the rest of the samples. When these genes were used to cluster the samples, a) two major branches were produced. Seventeen sample pairs were placed next to each other (small black brackets), but three were not (larger green brackets). b)f) Based on the four clusters of genes in the expression profiles, the samples were clustered into four main types: basal cell-like (orange); Erb-B2 positive (pink); normal breast tissue-like (green); luminal epithelial and estrogen receptor positive (blue). Three of the samples did not fit into any of these four categories.

There are at least 5 Different Breast Cancers49Figure 7.7 Cluster analysis using the intrinsic gene subset.The 496 genes were highly coordinated in each pair of before-and-after chemotherapy samples, but differed in the rest of the samples. When these genes were used to cluster the samples, a) two major branches were produced. Seventeen sample pairs were placed next to each other (small black brackets), but three were not (larger green brackets). b)f) Based on the four clusters of genes in the expression profiles, the samples were clustered into four main types: basal cell-like (orange); Erb-B2 positive (pink); normal breast tissue-like (green); luminal epithelial and estrogen receptor positive (blue). Three of the samples did not fit into any of these four categories.

No such thing as THE cure for cancer.50Figure 7.7 Cluster analysis using the intrinsic gene subset.The 496 genes were highly coordinated in each pair of before-and-after chemotherapy samples, but differed in the rest of the samples. When these genes were used to cluster the samples, a) two major branches were produced. Seventeen sample pairs were placed next to each other (small black brackets), but three were not (larger green brackets). b)f) Based on the four clusters of genes in the expression profiles, the samples were clustered into four main types: basal cell-like (orange); Erb-B2 positive (pink); normal breast tissue-like (green); luminal epithelial and estrogen receptor positive (blue). Three of the samples did not fit into any of these four categories.

The Discovery of Iressanon-small cell lung cancer

Success from Failure: Iressa

Cancer Treatment from Natures OdditiesNaked Mole Rat (NMR)

Face That Only A Mother Could Love

Face That Only A Mother Could Love?Cancer Treatment from Natures Oddities

lives 40 yearsnever has cancerlives 4 yearstumors commonhigh-molecular-mass hyaluronan (HA),

What prevents cancer in naked mole rats?

5 times bigger than human HAlower degradation rate than humanWhat prevents cancer in naked mole rats?

HA Over Produced in NMR Tissues

HA Over Produced in NMR Tissues

HA Over Produced in NMR TissuesHA Over Produced in NMR Tissues

relative amount of HABlock HA Tumors Form

neg. controlcarcinogen 1carcinogen 2carcinogen 3mouse cellsNMRNMR - HA

neg. controlcarcinogen 1carcinogen 2carcinogen 3NMRNMR - HAmouse cellsBlock HA Tumors Form

neg. controlcarcinogen 1carcinogen 2carcinogen 3NMR cellsNMR - HAmouse cellsBlock HA Tumors Form

neg. controlcarcinogen 1carcinogen 2carcinogen 3NMR no HAmouse cellsNMR cellsBlock HA Tumors FormInject Mice with NMR Cells +/- HA

tumorsNMR cellsNMR cells ++ HAaseNMR cells HA not madeinjectiontumorsrecipientmouse tumorInject Mice with NMR Cells +/- HA

NMR cellsNMR cells ++ HAaseNMR cells HA not madeinjectiontumorsrecipientmouse tumorInject Mice with NMR Cells +/- HA

NMR cellsNMR cells ++ HAaseNMR cells HA not madeinjectiontumorsrecipientmouse tumorInject Mice with NMR Cells +/- HA

NMR cellsNMR cells ++ HAaseNMR cells HA not madeinjectiontumorsrecipientmouse tumorInject Mice with NMR Cells +/- HA

NMR cellsNMR cells ++ HAaseNMR cells HA not madeinjectiontumorsrecipientmouse tumorInject Mice with NMR Cells +/- HA

NMR cellsNMR cells ++ HAaseNMR cells HA not madeinjectiontumorsrecipientmouse tumorInject Mice with NMR Cells +/- HA

NMR cellsNMR cells ++ HAaseNMR cells HA not madeinjectiontumorsrecipientmouse tumorInject Mice with NMR Cells +/- HA

NMR cellsNMR cells ++ HAasemouse tumorNMR cells HA not madeinjectiontumorsrecipient

Cancer Treatment from Natures Oddities

Remember this next time Congress makes fun of scientists studying odd creatures.Human Body Is Mostly Non-Human50 trillion human cells

50 trillion human cells500 trillion non-human cells

Human Body Is Mostly Non-Human50 trillion human cells500 trillion non-human cells10% human, 90% bacterial

Human Body Is Mostly Non-Human

Define the Human Microbiome27 sites 9 adults 4 occasions

each samplecolored by locationclustered by similar speciesDefine the Human Microbiome

Human Microbiome by Location

Human Microbiome by Locationhair and head skin similar

Human Microbiome by Locationhands and arms similar, L and R pairings

Human Microbiome by Locationmoist skin

Human Microbiome by Locationother moist places

Two Types of Skin Habitatstwo types of skin bacteria

Human Microbiome by Location

mouth and gut are really different from skin, bot as diverse

Human Microbiome Variationsmore variation between habitats than withinmore variation between people thane withinmore variation across time than on the same dayHuman Microbiome Variations

most diversity on skin than in oral cavity any time

Microbiome Transplant Experimentscompare to transplantscompare to nativesforeare did not change much after transplanting

Human Microbiome Variationscompare to transplantscompare to natives

forehead reverted to forehead microbiome

Human Microbiome Variationsforearm sustains transplantsforehead replaces transplantsHuman Microbiomemany skin sites more diversity than gut high-diversity skin sites forearmpalmindex fingerback kneesole of the foot Myth Busters: dogs mouth is really cleanMyth Busters: dogs mouth is really clean

mice fed low doses of antibiotics long term

15% more body fat than the control mice,Medical Uses of Microbiome

follow your nosesequenced bacteria in stool samples humans55 were thin122 overweight or obeseMedical Uses of Microbiome

obese people lack bacterial diversity in gutsantibiotics reduce microbiome diversityMedical Uses of Microbiome

obese people lack bacterial diversity in gutsmissing microbes methane producers,the carbon that does not get out as gas could be incorporated as fat.Medical Uses of MicrobiomeUsing six species, predict obese 80% of the time

Predictions based on genetics only 58% of the timeMedical Uses of MicrobiomeGuess what is being tested nowGuess what is being tested nowfecal transplantsThanks to my ColleaguesLaurie HeyerLeland Taylor 12supported by James G. Martin Genomics ProgramDavidson CollegeBlues390064.53