Transcript
Heading for full solution to Now Generation Informatics
BGI-Shenzhen
Sep 19, 2011
Nothing in biology makes sense except in
light of evolution
Theodosius Dobzhansky
“Tree” type of thinking of Genomics
They are different, they are also related
What is the scope of bioinformatics?
• Bioinformatics is to understand the tree of life.
• Bioinformatics will:
– Draw trees (basic information)
– Map information on trees (association/cause-effect)
– Show the trees (visualizations, databases, clouds)
Mission 1: Tree of Species
A set of different genes (sequence) made different forms of life
Mission 1: Tree of Species
• Draw– De novo genome assembly
– Multiple sequence mapping and alignment
– Phylogenic tree construction
• Map– In-depth Annotation
– Comparative genomicss
• Show– Genome browsers
Dinner
kung pao chicken
Peking Duck
Mapo toufu
Cabbagecucumber
oyster
“taste good, sequence it!”
Factory
Silk and silkworm Oil and castor bean
Cloth and cotton
“Useful, sequence it!”
Zoo
Panda Polar bear and Penguin Antelope
“look cute, sequence it!”
Misson 2: Tree of Individuals
A set of different variations (sequence) made different individuals/cells of Human
An Evolutionary perspective
The oldest human alleles originated in Africa well before the diasporas ofmodern humans 50,000 – 60,000 years ago.
These oldest alleles are common in all populations worldwide.
Approximately 90% of the variability in allele frequencies is of this sort.
From Mary-Claire King
• International project to construct a next generation baseline data set for human genetics
– Sequence level HapMap, an order of magnitude deeper
– Consortium with multiple centres, platforms, funders
• Aims
– Find >95% accessible SNPs at allele frequencies above 1%, down towards 0.1% in coding regions
– Genotype them and place on haplotype backgrounds
– Also discover and characterize indels, structural variants
An Evolutionary perspective
From Mary-Claire King
• Germlinede novo substitution rate =~ 1 x10-8 per generation
• Somatic/LCL substitution rate = 7-12x higher than germline rate
• Male mutation rate ~7x higher than female mutation rate
From 1000G Project
Development of agriculture in the past 10,000 years and of urbanization and industrialization inthe past 700 years has led to rapid populations growth and therefore to the appearance of vastnumbers of new alleles, each individually rare and specific to one population or even to one family.
What’s the whole picture of genetic variants ?
Allele Frequency
50% 5% 0.5% 0.05%Rarer Alleles
Stronger Effects
Common Alleles
Less Effects
Very Rare Alleles
Strongest Effects
Common/rare Disease Mendelian
Disease
Eg: MC4R, ABCA1 1q21.1 in SCZ
Eg: CFTR delta 508 PCSK9
C679X
Billion Genomes
Project
Personal genomics with phenotype information
• The history of silkworm domestication
D DomesticatedW wild
Silkworm domestication historySilkworm phylogenetic tree
Published in Science 16 Oct.
•relationship is not simply follow the geographic distribution which reflect gene-flow and other population level processes related to human activities such as ancient commercial trade• domestication event lead to a 90% reduction in effective population size during the initial bottleneck
from Anderssonand Georges, Nature Reviews of Genetic5: 202-212 (2004)
selective sweep: inheritance of regions around adaptive alleles
extent of selective sweep for domestication in MAIZE: tb1 locus (60 to 90-kb) (Clark et al. 2004), Y1 locus (about 600-kb) (Palaisa et al. 2004)
Domestication• Genome variation during silkworm domestication
Published in Science 16 Oct.
354 candidate domesticated genes
159 tissue-specific expressed (silk gland, midgut, testis)
50 Tibetan’s and 40 Han’s exomes has been sequenced
Function further validated in-Association with blood hemoglobin level-Expression level difference in placenta
EPAS1: endothelial Per-Arnt-Sim(PAS) domain protein 1
The gene (EPAS1) showing strongest selection signal (up to 80% frequency change in allele distribution), Han: 9%; Tibetan: 87%
The signal of selection
Your Micro-Environment, Your other genome?
PCA analysis for 85 Danish samples (based on gene profiling)
BMI data
Gene level
Misson 2. Tree of Individuals
• Draw
– (Complete spectrum of) variation identification
– Population frequencies and spectrums
• Map
– Selection and evolution
– Phenotypic traits
– Intermediate phenotypes
Misson 3: Tree of Cells
• Cell lineages are characterized by single biological levels and their inter-correlations.
+ : cancer*: normal*:cells possibly mixed （from tumor, but clustered to normal cells）
On DNADifferentiate the cancer and normal cells by PCA analysis
ET AML
BTCC
these cancers are really heterogeneous.
ET AML
Phylogenetic trees clearly show subpopulations in ET and AML cancers
Acute Myeloid LeukemiaEssential Thrombocythemia
Consensus Tree
Inferring key genes in AML (a typical heterozygous cancer)
Key Gene?
Key Gene for sub-pop?
G1~G6: different subpopulations from AML cancer
MLL: myeloid/lymphoid or mixed-lineage leukemia, recurrent translocations in acute leukemias that may be characterized as acute myeloid leukemia (AML; MIM 601626), acute lymphoblastic leukemia (ALL), or mixed lineage (biphenotypic) leukemia (MLL).
Key genes for AML
Inferring key genes in AML (a typical heterozygous cancer)
G1~G6: different subpopulations from AML cancer
LILRA1: leukocyte immunoglobulin-like receptor
G1~G6: different subpopulations from AML cancer
CTNNA1:Leukocyte transendothelial migration; Pathways in cancer
Inferring key genes in AML (a typical heterozygous cancer)
G1~G6: different subpopulations from AML cancer
CTSS: cathepsin
Inferring key genes in AML (a typical heterozygous cancer)
G1~G6: different subpopulations from AML cancer
PPP2R1A: TGF-beta signaling pathway
Inferring key genes in AML (a typical heterozygous cancer)
G1~G6: different subpopulations from AML cancer
DIAPH1: Focal adhesion; Regulation of actin cytoskeleton
Inferring key genes in AML (a typical heterozygous cancer)
G1~G6: different subpopulations from AML cancer
LILRA1: leukocyte immunoglobulin-like receptor
Inferring key genes in AML (a typical heterozygous cancer)
3. Tree of cells
• Draw
– Single-cell information acquisition technologies
• Map
– Single-cell metrics measurement technologies
Integrating DNA variation, molecular traits, and phenotypes to construct causal gene networks
Gene works in a network!
Finally: Where are the papers?
• On what paper you draw and map and show?– It is harder and harder to find a
platform efficient enough• Sample house
• High-throughput biology
• Capable computing system with high I/O performance
• Interlinked database and standardized formats
• Bioinformatics workflows to perform in silico analysis on data
0.1
1
10
100
1000
10000
2007 2008 2009 2010
Storage (Tb)
CPU power (TFLops)
Data generation (Tb)
Making data PUBLIC!
• Does not mean making data downloadable in theory
• Does mean the public could make use of data
• New types of databases with operations to the data are required
• New academic credit system to motivate high-quality easy-to-access datasets.
http://climb.genomics.cnhttp://www.gigasciencejournal.com
Acknowledgements
• Great International Efforts
– The Genome 10K Consortium
– The 1000 Genomes Project Consortium
– The 1000 Plant Genomes Project Consortium
– The 5000 insects Project Consortium (pending)
• BGI Initiatives and collaboration framework
– The 1000 Plant and Animal Genomes Project
– The 10K Microbial Genomes Project
http://ldl.genomics.org.cn
Acknowledgements
• Prof. Rasmus Nielson’s lab in UC Berkeley and in University of Copenhagen
• Prof. Richard Durbin’s lab in Wellcome Trust Sanger Insititute
• Prof. Tak-Wah Lam and Siu-Ming Yiu’s lab in Department of Computer Sciences, Hong Kong University
• Dr. Heng Li in Broad Insititute
• …
top related