Sequence Logos

The total height of each stack is determined by information content, measured in bits. At a given position i, the information content is calculated as:

R_i = log₂(4) - (H_i + e_n)

• log₂(4) = 2 bits: The maximum possible uncertainty for 4 nucleotides
• H_i: The Shannon entropy of the observed nucleotide distribution
• e_n: A small-sample correction factor to prevent overestimation

A perfectly conserved position (e.g., 100% Adenine) has 2 bits of information, while a completely random position has 0 bits.

Within each position, the height of each individual letter is proportional to its observed frequency multiplied by the total information content:

height(n) = f_n × R_i

• f_n: The relative frequency of nucleotide n at that position
• R_i: The total information content at position i

This ensures that both conservation (total stack height) and relative preference (individual letter height) are simultaneously visualized. A position with equal A and G frequencies will show both letters at equal heights.

Standard nucleotide color schemes provide immediate chemical group recognition:

• Adenine (A): Green — represents the purine base with an amino group
• Cytosine (C): Blue — represents the pyrimidine base with an amino group
• Guanine (G): Yellow/Orange — represents the purine base with a keto group
• Thymine (T): Red — represents the pyrimidine base with a keto group
• Uracil (U): Red — used in RNA logos, replacing Thymine

This amino-keto color grouping allows researchers to instantly distinguish between chemically similar nucleotides and identify functional substitutions.

Transcription factors bind DNA in a strand-agnostic manner, meaning a binding site on the forward strand is functionally equivalent to its reverse complement on the reverse strand. Sequence logos typically represent this by:

• Displaying the consensus orientation determined by the highest-scoring PWM match
• Using palindromic symmetry in logos for homodimeric transcription factors
• Applying reverse complement augmentation during motif discovery to merge counts from both strands

This property is critical for accurate genome-wide scanning, as binding events occur on both DNA strands.

Not all positions in a binding site are equally constrained. Sequence logos handle variable-length motifs through:

• Gap characters: Represented as empty or reduced-height positions where insertions or deletions occur in the alignment
• IUPAC ambiguity codes: Positions where multiple nucleotides are tolerated may display combined letters (e.g., R for A/G, Y for C/T)
• Low-information positions: Appear as short stacks with multiple small letters, indicating degenerate positions where the transcription factor has relaxed specificity

This visual encoding immediately highlights the core binding determinant positions versus flexible spacer regions.

A sequence logo is the direct visual transformation of a Position Weight Matrix (PWM) or Position Frequency Matrix (PFM). The conversion pipeline is:

1. PFM construction: Count nucleotide occurrences at each position from aligned binding sites
2. PWM calculation: Convert counts to log-odds scores relative to a background model (e.g., uniform 0.25 or genomic GC content)
3. Information content computation: Calculate R_i using Shannon entropy with small-sample correction
4. Logo rendering: Scale letter heights proportionally to f_n × R_i

Tools like WebLogo, ggseqlogo, and Logomaker automate this entire pipeline from aligned sequences to publication-ready graphics.