Skip to content

Chain Netting

prenderj edited this page Apr 14, 2021 · 7 revisions

Chain/net generation

The process of chain/netting takes the alignments generated in the previous stage, converts and combines them into a single chain file that is then refined:

  1. Alignments to psl: this largely depends on the output type of the aligner, and so there is not a unique tool to do this
  2. Generate chain: use axtChain to generate the chain file; the configuration changes depending on the distance of the genomes
  3. Filter out repetitive blocks (ChainAntiRepeat)
  4. Remove chains that can't be netted, and creates the net file (chainPreNet/chainNet)
  5. Add synteny info to the net file using netSyntenic (optional, can be turned off by --no_netsynt; recommended for medium and distantly related genomes)
  6. Subset the chain using the net file (chainNetSubset)

Pre-sets and custom configuration

The software uses several presets for the generation of the chain/net files.

If you use aligners like lastz and minimap2, it will use netSyntenic to refine the results, whereas using blat and GSAlign won't use it.

Another important factor is the generation of the chain file through axtChain. The software uses a series of presets, based on the aligner and distance configuration provided:

Preset axtChain
near -minScore=5000 -linearGap=medium
medium -minScore=3000 -linearGap=medium
far -minScore=5000 -linearGap=loose
primate -minScore=5000 -linearGap=medium
general -minScore=3000 -linearGap=medium
balanced -minScore=5000 -linearGap=medium
same -minScore=5000 -linearGap=medium

It is possible to switch off netSyntenic by using the --no_netsynt flag. It is also possible to specify custom axtChain parameters using the flag --chainCustom. The example below runs the workflow using the GRCh37 and GRCh38 genomes using lastz, with the near configuration for the alignments, and a custom axtChain configuration

nextflow run evotools/nf-LO --igenome_source GRCh37 \
   --igenome_target GRCh38 \
   --no_netsynt \
   --chainCustom '-minScore=5000 -linearGap=medium'

Clone this wiki locally