ReverseEngineeringofRegulatoryNetworks逆向工程的调控网络

1、Reverse Engineering of Regulatory NetworksKa-Lok NgDepartment of BioinformaticsAsia UniversityContents Introduction Gene regulatory network (GRN) The steady-state approach - time-series modelIntroduction - Gene regulatory network (GRN) One gene can affect the expression of another gene by binding of

2、 the gene product of one gene to the promoter region of another gene > 2 genes, we refer to the regulatory network as the regulatory interactions between the genes Given a larger number measurements of the expression levels of a number of genes, we should be able to model or reverse engineer the

3、regulatory network that controls their expression level. 2 approaches T the time-series and steady-state approaches Expression level of a gene at a certain time point Xj(t) can be modeled as some function of the expression levels of all other genes at all previous time points 斗化-1)， where i may of m

4、ay not equal / if i = j that meims self regulateN勺二ZX用(i)二where r-7- is a weight factor representing how gene i affects gene j, that is, i 9 j (activate) or z -I y (inhibit), positively or negativelyProblem!Many more genes > number of time pointsSuppose there are g genes g2 possible connections a

5、mong them (for instance, g 二 4 T16 possibilities, including self-regulation)There are g(g+1 )/2 possible interactionsIt is called one has a dimensionality problemA possible solution is cluster the genes that have similar behavior into gene clustersExampleAt t =0, gene c is induced, at t = 1,2,3,4 we

6、 follow the expression level of gene c and three other genes, a, b and dDeduce a GRN from the time-series data (5 time points).Time-seris experiment with four genesgene c T, but genes (a, b, d) gene c represses gene a ?1, 0, 1 今 inhibit, no interaction, activateGenetime01234a00-11b00011c01111d000014

7、 genes T 16 possible regulation relationsj = g (°)+m (°)+j +么(°)d g = j o = G (°) + (0) + J (0) + G (0) no_ conclusion厂 二1, 0, 1 inhibit, no interaction, activateGeneXtime01234a0o111b000-1-1c01111d0（）001bQ = 4） = rahu（t = 3） + rbibb（t = 3） + rCibc（t = 3） + rihbcl （t = 3）_K1）+ 皿（T

8、）+ S（l）+ 加（°）亠 T = T 加=> % at _t 3,2,1-1 = G（T）+ 加（°）+ G + 加（°）»1 二 一G + J 亠 N0 = a（°）+ U（°）+ G 0）+ 加（°）= S = 0 <jJ°二加（°）+加 （0） + /;方（0） + 巧/少（0） => no _ conclusionGeneXtime01234a00-1-11B000-11C01111d00001gene _cat_t-Ac(t = 4) = ra cu(t 二 3) + rh

9、 cb(t = 3) + rc cc(t = 3) + rd Ccl(f = 3)n)+ % (-1) + rc c (1) + rl c (0)1 = -rh c +1 z； c = 0at _t 3,2,11 = ra.c (T) + rh,c (°) + re,c + G (°) = 1 二 ra,c + 1= 01 = ra.c (0) + rb,c (°) + rc.e (0 + rd.c (0) e.c = 1(J £>1 = ra.c(0)+ rb.c(0) + rc c (0) + rd c (0) no _ conclusion-

10、|dgene _dat _t = 4Genetime01234a00-1-1-1B000-11C01111d0000-1a_bJ力d(f 二 4) = ra da(J = 3) + rb dbt = 3) +/= 3) + rd dd(f = 3)_=加(_1)广rh.d(-1)+ rc.d(I) + rd,d(OU i = -rbd 3 rhd = 1 ut _t = 3,2,10 =加(T)+m(o)+j +m(o)= o 二-加亠加=o0 二 ra,d(°)+ m(°)+ rc.d(0 + rd.d(°)3 id = °0 = ra d (0) + rb d (0) + rc d (0) + rd d (0) 3 no _ c