identification of associated transcription factors in
play

Identification of associated transcription factors in promoters and - PowerPoint PPT Presentation

Identification of associated transcription factors in promoters and their related enhancer regions Cornelia Meckbach Institute of Bioinformatics March 8, 2018 Cornelia Meckbach (Inst. of Bioinf.) Associated TFs in enhancer and promoters March


  1. Identification of associated transcription factors in promoters and their related enhancer regions Cornelia Meckbach Institute of Bioinformatics March 8, 2018 Cornelia Meckbach (Inst. of Bioinf.) Associated TFs in enhancer and promoters March 8, 2018 1 / 20

  2. Motivation Transcription factors (TFs) on paired enhancer and promoter regions are associated if they are involved in the pairing process. ⇒ Identification of associated TFs on enhancer and related promoter regions based on their transcription factor binding sites (TFBSs). Inspired by: Wong, KC (2017). MotifHyades: expectation maximization for de novo DNA motif pair discovery on paired sequences . Bioinformatics, Cornelia Meckbach (Inst. of Bioinf.) Associated TFs in enhancer and promoters March 8, 2018 2 / 20

  3. Mutual information Identification of associated TFs of promoter-enhancer pairings (PEPs) using mutual information (MI) → Two TFs are associated with each other if their binding behavior is in dependence of each other. Cornelia Meckbach (Inst. of Bioinf.) Associated TFs in enhancer and promoters March 8, 2018 3 / 20

  4. Mutual information Consider a set of experimentally validated PEPs for a cell line (e.g. by ChIA-PET) Predict all TFBSs of the underlying sequences Calculate MI for a TFBS pair T E and T P Cornelia Meckbach (Inst. of Bioinf.) Associated TFs in enhancer and promoters March 8, 2018 4 / 20

  5. Multivariate mutual information How much information contains TFBS T E about T P by considering the interaction type (label). Cornelia Meckbach (Inst. of Bioinf.) Associated TFs in enhancer and promoters March 8, 2018 5 / 20

  6. Workflow Cornelia Meckbach (Inst. of Bioinf.) Associated TFs in enhancer and promoters March 8, 2018 6 / 20

  7. Information theoretic measures Multivariate mutual information I( T E ; T P ;L) Cornelia Meckbach (Inst. of Bioinf.) Associated TFs in enhancer and promoters March 8, 2018 7 / 20

  8. Information theoretic measures Multivariate mutual information I( T E ; T P ;L) Mutual information of joint T E T P with L I( T E , T P ;L) Cornelia Meckbach (Inst. of Bioinf.) Associated TFs in enhancer and promoters March 8, 2018 8 / 20

  9. Information theoretic measures Multivariate mutual information I( T E ; T P ;L) Mutual information of joint T E T P with L I( T E , T P ;L) Conditional mutual information I( T E ; T P | L) Cornelia Meckbach (Inst. of Bioinf.) Associated TFs in enhancer and promoters March 8, 2018 9 / 20

  10. Information theoretic measures Multivariate mutual information I( T E ; T P ;L) Mutual information of joint T E T P with L I( T E , T P ;L) Conditional mutual information I( T E ; T P | L) Dual total correlation DTC( T E ; T P ;L) Cornelia Meckbach (Inst. of Bioinf.) Associated TFs in enhancer and promoters March 8, 2018 10 / 20

  11. Synthetic example I Synthetic TFBS-sequence matrix: An entry f ij in the matrix is the frequency of TFBS T j in sequence i . One row corresponds to a PEP The label column indicates the pairing type (true/false pair) Cornelia Meckbach (Inst. of Bioinf.) Associated TFs in enhancer and promoters March 8, 2018 11 / 20

  12. Synthetic example I � Perfect associated TFBS pair 1 Cornelia Meckbach (Inst. of Bioinf.) Associated TFs in enhancer and promoters March 8, 2018 12 / 20

  13. Synthetic example I � Perfect associated TFBS pair 1 � Associated TFBS pair in true PEPs 2 Cornelia Meckbach (Inst. of Bioinf.) Associated TFs in enhancer and promoters March 8, 2018 13 / 20

  14. Synthetic example I � Perfect associated TFBS pair 1 � Associated TFBS pair in true PEPs 2 � Non-associated TFBS pair 3 Cornelia Meckbach (Inst. of Bioinf.) Associated TFs in enhancer and promoters March 8, 2018 14 / 20

  15. Synthetic example I :Result � Perfect associated TFBS pair 1 � Associated TFBS pair in true PEPs 2 � Non-associated TFBS pair 3 Table: Results for different measures for synthetic example I. TFBS of TFBS of I( T E ; T P ;L) I( T E , T P ;L) I( T E ; T P | L) DTC( T E , T P ,L) enhancer promoter T E 1 T P 1 1.0 1.0 0 1.0 T E 2 T P 2 0.43 0.43 0.43 0.86 T E 3 T P 3 0 0.33 0.66 1.0 Cornelia Meckbach (Inst. of Bioinf.) Associated TFs in enhancer and promoters March 8, 2018 15 / 20

  16. Synthetic example II Use a given library of 166 PWMs to predict potential TFBSs in the sequences True PEPs: TFBSs V$IRF1 01 and V$USF 01 are randomly inserted 1 to 10 times in enhancer and promoter sequences False PEPs: Shuffled enhancer and promoter sequences Cornelia Meckbach (Inst. of Bioinf.) Associated TFs in enhancer and promoters March 8, 2018 16 / 20

  17. Synthetic example II: Results In total there are 27556 TFBS pairs → Ranks of the inserted pairs? Table: Ranking position of the inserted pairs. TFBS of TFBS of I( T E ; T P ;L) I( T E , T P ;L) I( T E ; T P | L) DTC( T E , T P ,L) enhancer promoter V$IRF1 01 V$USF 01 1 160 5563 125 V$IRF1 01 V$IRF1 01 2 161 6848 215 V$USF 01 V$IRF1 01 3 408 4309 60 V$USF 01 V$USF 01 4 396 3524 23 Cornelia Meckbach (Inst. of Bioinf.) Associated TFs in enhancer and promoters March 8, 2018 17 / 20

  18. Biolog. application: K562 cell line TFBS enhancer Logoplot of enhancer motif TFBS promoter Logoplot for promoter motif I( T E ; T P ;L) ( T E ) ( T P ) V$E2F Q6 01 V$CREB1 Q6 0.0144 V$CREB1 Q6 V$CREB1 Q6 01 0.0122 V$CREB1 Q6 V$CREB1 Q6 0.0115 V$E2F Q6 01 V$HOMEZ 01 0.0098 V$IK Q5 V$IK Q5 0.0095 V$IK Q5 V$HOMEZ 01 0.0094 V$E2F Q6 01 V$FAC1 01 0.0087 V$E2F Q6 01 V$HIF1A Q6 0.0084 V$CREB1 Q6 V$HOMEZ 01 0.0081 V$HIF1A Q6 V$E2F Q6 01 0.0075 Cornelia Meckbach (Inst. of Bioinf.) Associated TFs in enhancer and promoters March 8, 2018 18 / 20

  19. Summary Workflow to detect associated TFs on enhancer and promoter regions based on their binding sites Compared four different information theoretic measures on synthetic data sets Multivariate mutual information I( T E ; T P ;L) performs best on both sets Cornelia Meckbach (Inst. of Bioinf.) Associated TFs in enhancer and promoters March 8, 2018 19 / 20

  20. Thanks People Edgar Wingender Mehmet G¨ ultas Martin Haubrock Sebastian Zeidler J¨ urgen D¨ onitz Rayan Daou Darius Wlochowitz Halima Alachram Doris Waldmann Torsten Sch¨ ops Malte Sahrhage Cornelia Meckbach (Inst. of Bioinf.) Associated TFs in enhancer and promoters March 8, 2018 20 / 20

Recommend


More recommend