Output description

ThorAxe has two command-line programs, transcript_query and thoraxe. The first downloads from Ensembl the inputs for the second.

transcript_query

transcript_query PAX6 -l homo_sapiens,mus_musculus,xenopus_tropicalis,danio_rerio

Running the previous line is going to create a PAX6 folder with an Ensembl subfolder. That last folder contains four files with information downloaded from Ensembl. In this case, the information corresponds to the transcripts and exons from the human PAX6 gene and all their one-to-one orthologs in the indicated species:

PAX6
└── Ensembl
    ├── ensembl_version.csv
    ├── exonstable.tsv
    ├── sequences.fasta
    ├── tree.nh
    └── tsl.csv

The ensembl_version.csv file contains a table with information about the download date and hour together with the Ensembl Genomes version.

sequences.fasta contains the CDS sequence of each downloaded exon. You can find the used genome assembly in the sequence name/description. For example:

>homo_sapiens:MAPK8 ENSE00003837028 chromosome:GRCh38:10:48306753:48306821:1
CGGCGACCACCCCGGACGGCCCCTGTCCCCGCTGGCGGGCTTCCCTGTCGCCGTTCGCTGCGCTGCCGG

It’s the entry for MAPK8 exon ENSE00003837028 from chromosome 10 in the assembly GRCh38.

thoraxe

We can run thoraxe in the PAX6 folder that transcript_query created with the Ensembl data:

thoraxe -i PAX6 --phylosofs --plot_chimerics

ThorAxe is going to look for the Ensembl folder with the input data and is going to create a thoraxe folder containing the final and intermediate results of the study. The –phylosofs and –plot_chimerics arguments are optionals.

PAX6
├── Ensembl
│   ├── ensembl_version.csv
│   ├── exonstable.tsv
│   ├── sequences.fasta
│   ├── tree.nh
│   └── tsl.csv
└── thoraxe
    ├── _intermediate
    │   ├── chimeric_alignment_1.fasta
    │   ⋮
    │   ├── cluster_data.js
    │   ├── cluster_plots.html
    │   ├── gene_ids_1.txt
    │   ⋮
    │   ├── msa_matrix_1.txt
    │   ⋮
    │   ├── subexon_table.csv
    │   ├── subexon_table_1.csv
    │   ⋮
    │   └── transcript_table.csv
    ├── ases_table.csv
    ├── msa
    │   ├── msa_s_exon_10_0.fasta
    │   ⋮
    │   └── msa_s_exon_9_0.fasta
    ├── path_table.csv
    ├── phylosofs
    │   ├── s_exons.tsv
    │   ├── transcripts.pir
    │   ├── transcripts.txt
    │   └── tree.nh
    ├── s_exon_table.csv
    └── splice_graph.gml

The msa folder has a multiple sequence alignment for each s-exon, e.g. msa_s_exon_10_0.fasta. This MSAs can be easily used as a seed to look for homologous sequences in different databases by using hmmsearch.

Tables

ThorAxe outputs three tidy (denormalized) tables rather than a normalized database: s_exon_table.csv, path_table.csv and ases_table.csv. These tables are comma-separated files easily read by R, Python (using pandas) or Julia (using CSV and DataFrames) among other software tools. When a cell should include a list of elements, ThorAxe uses by default the character / as the delimiter between the list elements. The stored list can easily re-create by using strsplit in R, the split string method in Python or the split function in Julia.

S-exon table

The s_exon_table.csv file has the tidy (denormalized) output data for the orthologous exonic regions (s-exons). Each row is an observation of an s-exon in a particular transcript. Rows are in order, following the s-exon rank. That means that the concatenation of the s-exon sequences for each transcript gives the protein isoform sequence.

Path table

The path_table.csv contains information about each transcript as a path in the splice graph. The first row indicates the canonical path. The canonical path is used to detect alternative splicing events automatically, and it is the one present in the largest number of genes/species with the highest minimal value of transcript_weighted_conservation, and it is the longest. The alternative splicing events are stored in the ases_table.csv file where the canonical path and the alternative are indicated, together with conservation (number of genes showing the path) information.

Table of Alternative Splicing Events

ThorAxe automatically detects alternative splicing events by comparing each transcript/path to the canonical path in the splice graph. The ases_table.csv file stores these detected events. The table identifies each event using the first two columns: CanonicalPath and AlternativePath. These columns store the list of nodes that forms the alternative subpaths in the canonical and alternative (non-canonical) path. First, ThorAxe classifies each event, respect to the canonical path, as insertion, deletion , fully_alternative, alternative_start, alternative_end or simply alternative. The ASE column stores this classification. The events, excepting insertions and deletions, are further classified according to their number of mutually exclusive s-exons pairs into: mutually_exclusive, partially_mutually_exclusive and not mutually exclusive (empty cell). The MutualExclusivity column stores this classification. The MutualExclusiveCanonical and MutualExclusiveAlternative columns stores two paired list indicating the mutually exclusive s-exon pairs. For example, an event having 15_0/15_1 in MutualExclusiveCanonical and 7_5/7_5 in MutualExclusiveAlternative, means that s-exon 15_0 is mutually exclusive with s-exon 7_5 and that 15_1 is also mutually exclusive with 7_5. Here, mutually exclusive means that there is no transcript in the path table showing both s-exons.

Splice graph

ThorAxe splice graph as s-exons, rather than exon, as nodes. Using s-exon allows us to represent transcripts from different species in the same splice graph. As a consequence, the edges indicate the connectivity between these exonic regions and not necessarily junctions between genomic exons. ThorAxe splice graph has conservation information, where conservation means the fraction or transcript weighted fraction of species showing a particular node (s-exon) or connection. The splice_graph.gml file stores this splice graph of s-exons using the human-readable Graph Modelling Language (GML) format. GML is a rich format, and we used it to store useful metadata for nodes and edges:

Metadata	Description
label	The s-exon ID (only for nodes).
transcript_fraction	It is the number of transcripts showing that node/edge over the total number of transcripts, taking into account all the genes.
conservation	Fraction of genes showing that node/edge.
transcript_weighted_conservation	As conservation but each gene is weighted using the fraction of transcripts in that gene showing that edge (only for edges).
genes	List of genes, separated by commas, showing that node/edge.
transcripts	List of transcripts, separated by commas, showing that node/edge.

Note

If ThorAxe’s transcript_query is run with –orthology 1:1 (default), the number of genes is identical to the number of species in the dataset.

phylosofs

The phylosofs folder has the needed inputs for the structural and molecular modelling pipelines of PhyloSofS. It is only generated when the –phylosofs optional argument is used. These PhyloSofS’s files use a single unicode character to represent each s-exon. The mapping between the id for PhyloSofS and the one of ThorAxe (ExonCluster_ChimericBlock) is in the s_exons.tsv file.

The transcripts.pir has the annotated sequence using the PIR format.

The list of transcripts for each gene is in transcripts.txt

Intermediate outputs

The _intermediate folder has intermediate files and states from the ThorAxe pipeline. In particular, cluster_plots.html has the interactive plots of the chimeric alignments, with the possibility to show the constitutive value calculated for each subexon. This plot is only generated when the –plot_chimerics optional argument is indicated.

Warning

cluster_plots.html loads the cluster_data.js file that includes all the information needed to plot all the generated multiple sequence alignments using Plotly. It can freeze the tab when a lot of sequences/species are used or if the computer hasn’t enough resources available.