Supplementary data for "Epigenetic priors for identifying active transcription factor binding sites"

Gabriel Cuellar-Partida, Fabian A. Buske, Robert C. McLeay, Tom Whitington, William Stafford Noble and Timothy L. Bailey

Bioinformatics. 28:56-62, 2011.

---

Experiment 1 (Section 3.1)

• TF ChIP-seq data sets from Mouse ESC:
  1. ChIP-seq_mouse.tgz 1.3M
    ChIP-seq_mouse/$TF.bed
    BED files from ChIP-seq assays for the described transcription factors.
• Gold standard and results:
  1. Mouse_experiments.tgz 2.6G
    Mouse_experiments/$TF.Prior.Fil.chr#.txt
    Tab-separated columns:
  1. Chr
  2. Start position
  3. Strand
  4. PWM score (FIMO)
  5. P-value (FIMO)
  6. 1/-1 (Positive/Negative) (Gold standard)
  7. Log-posterior odds score, described in this paper using H3K4me3 Mouse ESC prior
  8. Filtering number (PWM score if passed threshold of 1, PWM score-100 otherwise)
  9. Filtering number (PWM score if passed threshold of 2, PWM score-100 otherwise)
  10. Filtering number (PWM score if passed threshold of 4, PWM score-100 otherwise)
  11. Filtering number (PWM score if passed threshold of 8, PWM score-100 otherwise)
  12. Filtering number (PWM score if passed threshold of 16, PWM score-100 otherwise)
  13. Actual H3K4me3 density (-1, if it couldn't be determined (repetitive region)

Experiment 2 (Section 3.2)

• TF ChIP-seq data sets from K562 cell line:
  1. ChIP-seq.tgz 2.5M
    ChIP-seq/ $TF.bed
    Bed file of ChIP-seq assays for the described transcription factors.
• The next set of files correspond to those used to create the priors (those without prior suffix), and the priors (prior suffix). The files are on wiggle format.
  1. DNase.tgz 27M
  2. DNasePrior.tgz 47M
  3. H3K4me1.tgz 49M
  4. H3k4me1Prior.tgz 75M
  5. H3K4me3.tgz 44M
  6. H3K4me3Prior.tgz 69M
  7. H3K9ac.tgz 50M
  8. H3K9acPrior.tgz 78M
  9. H3K27ac.tgz 39M
  10. H3K27acPrior.tgz 63M
• Gold standard and results
  1. Human_experiments.tgz 7.5G
    Human_experiments/$TF_27ac9ac.me3.me1.DNase.Combined.chr#.txt
    Tab-separated columns:
  1. Chr#
  2. Start position
  3. Strand
  4. PWM score (FIMO)
  5. P-value (FIMO)
  6. 1/-1 (Positive/Negative) (Gold standard)
  7. Log-posterior odds score, H3K27ac K562 prior
  8. Log-posterior odds score, H3K9ac K562 prior
  9. Log-posterior odds score, H3K4me3 K562 prior
  10. Log-posterior odds score, H3K4me1 K562 prior
  11. Log-posterior odds score, DNase HS K562 prior
  12. Log-posterior odds score, H3K4me3 + DNase HS K562 prior
  13. Log-posterior odds score, H3K27ac + DNase HS K562 prior
  14. Log-posterior odds score, H3K4me1 + H3K4me3 K562 prior
  15. Log-posterior odds score, H3K9ac + H3K27ac K562 prior
  16. Log-posterior odds score, H3K4me3 + H3K9ac + H3K27ac K562 prior
  17. Log-posterior odds score, H3K4me1 + H3K4me3 + H3K9ac + H3K27ac K562 prior

Experiment 3 (Section 3.3)

• Gold standard and results
  1. centipede.tar.gz 3.3M
    centipede/$TF.bed
    The format of the files is similar to the UCSC bed format, which each column.
    Note that columns 1-11 are from CENTIPEDE paper:
    Pique-Regi R*, Degner JF*, Pai AA, Gaffney DJ, Gilad Y, Pritchard JK. Accurate inference of transcription factor binding from DNA sequence and chromatin accessibility data. Genome Research. 2011. Jan 20.
  1. Chromosome
  2. Starting position of the motif (0-based) hg18
  3. End position of the motif (1-based) hg18
  4. Identifier for the PWM (Transfac, or Jaspar ID).
  5. Score for the PWM match (as described in the supplement log2 (likelihood ratio PWM model / simple 0.25 background model) )
  6. Strand of the match ("+" Forward, "-" Reverse)
  7. Posterior log odds of TF binding from the Centipede model
  8. Posterior probability of TF binding from the Centipede model
  9. Number of ChIPseq-reads surrounding the motif match
  10. Number of control ChIP-reads surrounding the motif match
  11. If inside a ChIP-seq peak called by Macs or Encode (see supplement for details)
      1 -- Inside a high confidence ChIP-seq peak (POSITIVE)
      0 -- Defined as ChIP-seq peak NEGATIVE
  12. DNase prior score
  13. DNase score

---

Please direct questions to Gabriel Cuellar-Partida (gcuellar@lcg.unam.mx), William Noble (william-noble@uw.edu) or Timothy Bailey (t.bailey@imb.uq.edu.au).