Massimo Vergassola CNRS, Institut Pasteur

Transcription

1 Massimo Vergassola CNRS, Institut Pasteur Dr. Massimo Vergassola, Institute Pasteur, Paris (KITP Bio Seminar 11/23/04) 1

2 ! "# $!% Bcd,nos,tor, dl,cad, Hb,Kr,tll,... Eve,ftz,h,... En,wg,dsh,... Dr. Massimo Vergassola, Institute Pasteur, Paris (KITP Bio Seminar 11/23/04) 2

3 &! % Dr. Massimo Vergassola, Institute Pasteur, Paris (KITP Bio Seminar 11/23/04) 3

4 ' # ( )"*+ +,% Expression profiles are a superposition of elementary patterns. Stretches of a few hundred bases in the vicinity of the regulated gene yield those patterns by a strongly combinatorial regulation of transcription. Proximal Eve Promoter Kb +1 # * Dr. Massimo Vergassola, Institute Pasteur, Paris (KITP Bio Seminar 11/23/04) 4

5 # -. / * Dr. Massimo Vergassola, Institute Pasteur, Paris (KITP Bio Seminar 11/23/04) 5

6 0 The size of the regions to explore calls for cheaper and faster clues on their location. Dr. Massimo Vergassola, Institute Pasteur, Paris (KITP Bio Seminar 11/23/04) 6

7 1 2 3 ChIP genome-wide data (Lee et al. Science, 298, 799, 2002) # genes # regulators 2! 0% Goal: maximize the likelihood to observe the input sequence given the list of bricks: TF s (as weight matrices w k ) and the background model (as transition probabilities of a Markov chain w 0 ). w i1 w i2 w i3 w i4 w i5 w i6 s 1 s 2 s 3 s 4 s 5 s 6 Tiling T P( T) = N( T) k= 1 p w ik m( s k w i k ) Dr. Massimo Vergassola, Institute Pasteur, Paris (KITP Bio Seminar 11/23/04) 7

8 2! 00% The competition among factors for overlapping sites is brought in by the partition function Z = P(T ) Maximize Z over the p w s to get the maximum likelihood Z max via the transfer matrix method Z (1, i) = N w k = 0 Z ( i l Score of the sequence R = log k ) p ( Z ) max wk Z back ( w ) m s ik k! 4 Dr. Massimo Vergassola, Institute Pasteur, Paris (KITP Bio Seminar 11/23/04) 8

9 $!"! #"! #! $ && % upstream construct Dr. Massimo Vergassola, Institute Pasteur, Paris (KITP Bio Seminar 11/23/04) 9

10 5 % hasa upstream region (regulated by CovR) 2' Dr. Massimo Vergassola, Institute Pasteur, Paris (KITP Bio Seminar 11/23/04) 10

11 ';*%,6-7 *8-7.7 **% *.7 < % -.7 <,97 ; -7 :,*%,, 97 8,7 &! 2,8%,, = Lf Gene Remarks Function 250 gbs0042 Transcriptome Similar to phosphoribosylamine-glycine ligase 250 gbs0358 Transcriptome Unknown 1000 gbs0493 Transcriptome Similar to unknown protein 1000 gbs0595 Pathogenicity island Similar to ABC transporter (ATP-binding protein) 1000 gbs1087 Transcriptome Highly repetitive peptidoglycan bound protein (LPTX motif) 1000 gbs1194 Transcriptome Similar to unknown protein 500 gbs1279 Transcriptome Similar to putative hydrolytic protein 1000 gbs1312 Pathogenicity island Exonuclease motif predicted by PFAM 250 gbs1506 Transcriptome Similar to glycerol (sugar)-3-phosphate transporter 1000 gbs1609 Transcriptome Unknown 100 gbs1919 Transcriptome Similar to neuramidinase 1000 gbs1972/ Pathogenicity island Similar to transcriptional regulator (phage related) /unknown gbs gbs2018 Transcriptome Putative peptidoglycan linked protein (LPTX motif) 1000 gbs2074 Pathogenicity island Unknown Dr. Massimo Vergassola, Institute Pasteur, Paris (KITP Bio Seminar 11/23/04) 11

12 '- ' +,*! CovS "#$% &' "$%( &' "$%) &' CovR CovS~P CovR~P, & " $$) * ")$ ++ "$#( '+& & :4 ' )# ( % Dr. Massimo Vergassola, Institute Pasteur, Paris (KITP Bio Seminar 11/23/04) 12

13 1 >?'."68;"*+++% 4 Dr. Massimo Vergassola, Institute Pasteur, Paris (KITP Bio Seminar 11/23/04) 13

14 "(!"A "B = C / ( D *+ +;% List of putative orthologs built by bi-directional best hits to reduce the role of paralogs. For each organism of comparison, evolutive ranks r i for proteins of S. cerevisiae obtained by ordering them in increasing order of the Blast e-value of the alignment to their best hit. Evolutive distance cliques of proteins. D= ( r i r j ) i< j 2 computed for Dr. Massimo Vergassola, Institute Pasteur, Paris (KITP Bio Seminar 11/23/04) 14

15 B E Small D values are indicative of multi-point co-evolutive correlations. D= ( r i r j ) i< j 2 r i evolutive ranks of the n proteins composing the clique. $ 4 # proteins interlinked # groups # proteins z-score 8.1 ± ± ± ± ± ± ± ± 0.1 probability # proteins interlinked # groups # proteins z-score 8.0 ± ± ± ± ± ± ± ± 0.1 probability Dr. Massimo Vergassola, Institute Pasteur, Paris (KITP Bio Seminar 11/23/04) 15

16 0! # proteins interlinked # groups # proteins z-score 8.0 ± ± ± ± 0.1 Transcription factors Normalized number of proteins All Essential S. Ghaemmaghami et al. Nature 425, 737 (2003) < >20 log (molecules per cell) : ' Abundant prots. ranked high Abundance rank abundance rank Evolutive divergence rank C. glabrata D. hansenii K. lactis Y. lipolytica Evolutive divergence rank Conserved prots. ranked first Same behavior found in Pál, C. et al. Genetics (2001) for the comparison with C. albicans based on transcript levels. Dr. Massimo Vergassola, Institute Pasteur, Paris (KITP Bio Seminar 11/23/04) 16

17 0 : )' U. Gaul, E. Siggia (Rockefeller Univ.); N. Rajewsky (New York Univ.) (BMC Bioinformatics, 3:30) P. Glaser, F. Kunst, T. Msadek, P. Trieu-Cuot (Molecular Microb., to appear) B. Dujon (Inst. Pasteur); A. Vespignani (Indiana Univ.) (Proteomics, to appear) Dr. Massimo Vergassola, Institute Pasteur, Paris (KITP Bio Seminar 11/23/04) 17