SyllabusResourcesDataInstructorsGitHub ↗

W1_L03

Phylogenetics 101 — Part B

↗ Slides📓 Lab Notes

Core Idea (bird's-eye view)

Lesson 02 was about trees as objects (Newick/Nexus). Lesson 03 is about the data that goes into building those trees: molecular sequences. The practical is short and format-focused (FASTA, PHYLIP, Nexus-DATA), but it sits on top of a heavy theoretical layer you must carry into the meeting:

  • Phylogenetics can use any heritable character, but molecular characters dominate because they are abundant, cheap, objective, standardized, and resolve both shallow and deep relationships.
  • Characters used must be: heritable, independent, informative, clearly coded, homologous, and ideally non-homoplasious.
  • Homology (similarity from shared ancestry) is the prerequisite. Homoplasy (convergence, reversal) is the enemy — it produces character conflict and misleads inference.
  • Genetic distance between two taxa = sum of changes along the two branches connecting them to their common ancestor: d(SP1,SP2) = (SP0 to SP1) + (SP0 to SP2).
  • Molecular clock hypothesis (Kimura, neutral theory): neutral mutations accumulate at a roughly constant rate, so older splits accumulate more differences. This is what allows dating divergences — but rates vary across genes, lineages, and time, so calibrations (fossils, known splits) are needed.
  • Historical anchors: Nuttall (1904, immunological distances), Cavalli-Sforza & Edwards (1963, first computed trees), Fitch & Margoliash (1967, first modern sequence-distance tree).

The practical itself is teaching you how molecular data is physically stored on disk so you can feed it into alignment and tree-building tools later in the course.

Inputs

The practical introduces three sequence file formats. The "input" here is raw sequence data — nucleotides (DNA/RNA) or amino acids — for multiple taxa.

1. FASTA

The simplest and most common format. Each sequence has a header line starting with > followed by the sequence on the next line(s).

>Taxon1
ATGCGTACGTAGCTAGCTACGATCG
>Taxon2
ATGCGTACGTAGCTAGATACGATCG
>Taxon3
ATGCGTACGTAGCTAGCTACGATCC

Key points:

  • Header line must start with >.
  • No length declaration, no strict columns. Very flexible.
  • Used everywhere — alignments, databases (NCBI, UniProt), tool inputs.

2. PHYLIP

A strict column-based format originally for the PHYLIP software package.

3 25
Taxon1    ATGCGTACGTAGCTAGCTACGATCG
Taxon2    ATGCGTACGTAGCTAGATACGATCG
Taxon3    ATGCGTACGTAGCTAGCTACGATCC

Key points:

  • First line: <number of sequences> <number of characters>.
  • Taxon names traditionally limited to 10 characters — longer names get truncated. This is a real source of bugs.
  • Two flavors exist: sequential (one taxon at a time) and interleaved (blocks of all taxa, repeating). The example above shows the interleaved style.
  • Less flexible than FASTA but still expected by many phylogenetics tools.

3. Nexus (DATA block)

The same Nexus container from Lesson 02, but here storing a sequence matrix instead of a tree.

#NEXUS
BEGIN DATA;
    DIMENSIONS NTAX=3 NCHAR=25;
    FORMAT DATATYPE=DNA MISSING=? GAP=-;
    MATRIX
        Taxon1    ATGCGTACGTAGCTAGCTACGATCG
        Taxon2    ATGCGTACGTAGCTAGATACGATCG
        Taxon3    ATGCGTACGTAGCTAGCTACGATCC
    ;
END;

Key points:

  • Block-based: BEGIN DATA; ... END;.
  • Declares dimensions (NTAX, NCHAR), datatype (DNA/RNA/protein), and the symbols used for missing data and gaps.
  • Most flexible — can store sequences, trees, character sets, and analysis commands all in one file. This is why programs like MrBayes and PAUP* use it natively.

Outputs

The practical does not "produce" anything computed — its outputs are conceptual:

  • The ability to recognize at a glance which format a file is in.
  • Understanding that the same sequence data can be written in any of these three formats and converted between them losslessly (modulo PHYLIP's name-length limit).
  • Knowing which format which tool expects — FASTA for aligners (mafft), PHYLIP for many tree builders (older iqtree, RAxML), Nexus for Bayesian tools (MrBayes) and tree storage.

Interpretations

  • Format choice is downstream-driven: you pick the format the next tool in your pipeline wants. Conversions are routine.
  • PHYLIP's 10-character name limit is the most common gotcha. If your taxa have long names, they get silently truncated, and downstream you may end up with duplicated or confused labels.
  • FASTA has no built-in concept of alignment: the sequences in a FASTA file may or may not be aligned. An "aligned FASTA" is just a FASTA where every sequence has the same length (gaps usually shown as -). This matters because phylogenetics needs aligned data, not raw sequences.
  • Nexus is a container, not a format: the same #NEXUS file can hold a DATA block (Lesson 03 practical) or a TREES block (Lesson 02 practical), or both at once. This is why Nexus shows up in both lessons.
  • Connecting back to theory: each column in an aligned matrix is a character; each cell is a character state. The whole matrix is the input to every algorithm in this course. The quality of phylogenetic inference is bounded by how well the columns represent homologous positions — which is the entire job of sequence alignment (Lesson 05).
  • On the molecular clock: the data in these files is what you measure distance on. Whether that distance reflects time depends on whether the rate of change has been constant — which the prof has flagged as usually NOT being the case in practice.

Lexicon to keep handy

  • homology / homoplasy / convergence / reversal
  • heritable, independent, informative, clear, homologous, non-homoplasious (the six character criteria)
  • genetic distance, molecular clock, neutral theory
  • FASTA, PHYLIP (sequential vs interleaved), Nexus DATA block
  • aligned vs unaligned sequences
  • NTAX, NCHAR, DATATYPE, GAP, MISSING

Possible Exam Questions

(to be filled in during the meeting)