Universal Sequence Logos

macbook.universal.logoThe new Universal Sequence Logo Tool produces publication-quality logos that integrate high-order sequence dependencies and corresponded positions for PWMs, PSAMs, Di-PSAMs, and pHMMs. Currently supports DNA, methylated-DNA, RNA, and methylated-RNA binding models. In addition, the logos include a Correspondence Panel that depicts corresponded columns with tied parameters and shared binding information. The logos also provide biophysical measurements including relative affinities (Kas) and changes in binding free energies (ddGs) for each sequence feature. In order to support shared, high-order positional dependencies in a symmetric motif, directionality in the dependencies is also supported. Lastly, we introduce an integrated XML schema for annotating these high-order sequence dependencies and corresponded positions for PWMs, PSAMs, Di-PSAMs, and pHMMs. Current functionality includes:

  1. Support for both Markov (probabilistic) models and relative affinity models:
    • Markov models: PWMs, Di-PWMs, and pHMMs
    • Affinity models: PSAMs, Di-PSAMs, and affinity-pHMMs
  2. Currently supports 4 different types of nucleic-acid-binding models:
    • DNA binding: consisting of adenines (A), cytosines (C), guanines (G), and thymines (T)
    • methylated-DNA binding: augments DNA binding with methylated cytosines (Z) and guanines (J) base-complemented with methylated cytosines
    • RNA binding: consisting of adenines (A), cytosines (C), guanines (G), and uracil (U)
    • methylated-RNA binding: augments RNA binding with methylated cytosines (Z) and guanines (J) base-complemented with methylated cytosines
  3. Correspondence panel (optional) – depicts corresponded columns with tied parameters and shared binding information.
  4. Biophysical metrics – relative affinities (Kas) and changes in binding free energy (ddG in Kcal/mol) for each sequence feature.
    • In the presence of positional dependencies, the relative affinity (Ka) and basis binding free energy (ddG) of the positional independence PSAM is adjusted so that the highest affinity binding site has total Ka and ddG of 1.0 and 0.0, respectively, after the corrections to the PSAM due to the positional dependencies.
  5. Dependency corrections to the independence model –models positional dependencies by augmenting the positional-independence Sequence Logo to include multiplicative affinity-correction terms for positions in the binding site where the positional-independence assumption breaks down.
    • Every higher-order feature is presented with its corresponding energetic affinity correction to the positional-independence model.
    • Three different classes of features are currently supported:
      • nearest-neighbor dinucleotide corrections
      • nearest-neighbor trinucleotide corrections
      • context-dependent nucleotide insertions
  6. Directional dependencies – dependencies at any position has either a forward or reverse direction
    • Allows for symmetric binding models with corresponded columns that have reversed dependencies
  7. Feature height scaling (optional) – scales the relative height of the features to represent their relative correctional magnitude.
    • The green rows represent sequence features that increase binding affinity.
    • The red rows represent sequence features that decrease binding affinity.
  8. Column width scaling (optional) – scales the widths of the columns for a pHMM model according to the transition rates or affinities.