Post-analysis of MIGMAP output
To summarize somatic hypermutations and generate clonotype trees run
java -Xmx8G -jar migmap.jar com.antigenomics.migmap.Analyze -S human -R IGH raji_R12.fastq raji_R12
This folder contains analysis results for IGH repertoire of hypermutating Raji cell line:
- Reads stored in
raji_R12.fastq.gz.. actually, those are not raw reads, but error-corrected assembled consensuses, see MIGEC.
- Clonotype table and binary output are stored in
Analysis of somatic hypermutation profile
raji_R12.shm.txt- mutation table
analyze_shm.Rmd- analysis template
The whole analysis template is stored in R markdown format, which means it can be loaded to Rstudio, customized and executed. Detailed explanation of all analysis steps and plots is embedded into the template. An example PDF output is shown in
analyze_shm.pdf (click here to view it).
out.net.txt- network graph
out.node.txt- node attributes
out.edge.txt- edge attributes
The analysis can be entirely performed in Cytoscape software as follows.
Import the network from tab-delimited files generated by MIGMAP
Node Table columnsand
Edge Table columnsrespectively under
Import Data as...
Specify parameter mapping
- Node size: continuous mapping for
- Node label: passthrough mapping for
- Node label color: discrete mapping for
cdr3.codeattribute. Then apply (right-click) a Mapping Value Generator. Clonotypes with close
cdr3.codevalues have similar CDR3 sequences.
- Target arrow shape: select an arrow, the graph is directed and built using parsimony principle.
- Edge color: continuous mapping for
replacement.ratioattribute. Note that it is not S:R ratio, but rather number of replacement mutation among all mutations. Only mutations that are present in target node, but not in its parent (i.e. the difference) are counted here.
- Width: continuous mapping for
shm.count.negattribute. The more hypermutations separate two clonotypes, the smaller is the edge width.
Edge weighted Spring Embedded Layout from the
Layout menu, use
shm.count.neg as parameter. An example cytoscape file with the network for Raji cell line that can be also used as a template is stored in this folder (
Under masterment The concept is a network showing each CDR3 with its children (edges are weighted by hypermutation load/S:R ratio) and edges between distinct CDR3 which are likely hypermutations. E.g. a single mutation within "N" region of CDR3 and consequent match of 5 "N" nucleotides between two different CDR3 tells that they are more likely a result of hypermutation than of independent V-D-J recombination.