Contributors: Hardip Patel, Paul Waters, Aurora Ruíz-Herrera, Arthur Georges, Jenny A.M. Graves

The process was automated using the preparegenomes.sh script. Basic steps were as follows.

1. Genome assemblies were downloaded from relevant sources (NCBI or DNAZoo). Check metadata/species.txt for details of species and download paths.
2. .2bit, .capsule, .sizes files were created.
3. Genome files were split into smaller regions (1Mb sequence size and ~5Mb of total sequence per file) without overlap. This step allows for embarassingly parallel lastz alignments to capitalise on large HPC facilities.

We used the following parameters for LastZ alignments:
K=2400 L=3000 Y=9400 H=2000 --ambiguous=iupac

List of species considered for analyses

Typical workflow

This workflow is automated using getMAF.sh script.

1. Fix LastZ alignments to assign genomic coordinates to sequence alignments as alignments were performed for 1Mb sub-sequences (fixlastz.pl).
2. Performing chaining of LastZ alignments (axtChain).
3. Sort chains (chainSort).
4. Generate prenet files from chain output (chainPreNet).
5. Perform netting of chains (chainNet).
6. Report nets of alignments against target genome (netSyntenic).
7. Report nets of alignments against query genome (netSyntenic).

Further steps that can be peformed but not performed as yet for this project. Commands are included in the getMAF.sh script, however, we have commented them out for now.

1. Convert nets to .axt format (netToaxt).
2. Sort .axt alignment files (axtSort).
3. Generate .maf format output for alignments (axtToMaf).

Credits

1. Workflow ideas for lastz alignments were borrowed from Daren Card (thanks mate) available here.

2. Lastz alignment parameters were obtained from several sources:

   • The bird genome alignment paper GitHub repo here.

   K=2400 L=3000 Y=9400 H=2000
   

   • Alignment sensitivity tests performed in Sharma et al, 2017 paper.

   K = 2400, L = 3000, Y = 9400, H = 2000 for placental mammals
   K = 2400, L = 3000, Y = 3400, H = 2000 for non-placental mammals
   K = 1500, L = 2500 and W = 5  to find co-linear alignments in the un-aligning regions that are flanked by local alignments (gaps in the chains)
   

   • Ensembl Compara LastZ pairwise alignment settings for the GitHub Repo here

   default => 'T=1 L=3000 H=2200 O=400 E=30 --ambiguous=iupac', # ensembl genomes settings
   # Vertebrata
   7742    => 'T=1 K=3000 L=3000 H=2200 O=400 E=30 --ambiguous=iupac',
   # Catarrhini, Sus, Carnivora, Triticeae
   9526    => 'T=1 K=5000 L=5000 H=3000 M=10 O=400 E=30 Q='
   9822    => 'T=1 K=5000 L=5000 H=3000 M=10 O=400 E=30 --ambiguous=iupac',
   33554   => 'T=1 K=5000 L=5000 H=3000 M=10 O=400 E=30 --ambiguous=iupac',
   147389  => 'T=1 K=5000 L=5000 H=3000 M=10 O=400 E=30 --ambiguous=iupac --identity=75..100',
   # Vigna, Solanaceae
   3913    => 'T=1 L=3000 H=2200 O=400 E=30 --ambiguous=iupac --matchcount=1000',
   4070    => 'T=1 L=3000 H=2200 O=400 E=30 --ambiguous=iupac --matchcount=1000',
   #4107    => 'K=5000 L=5000 H=3000 O=400 E=30 --ambiguous=iupac M=10 --notransition --step=20',
   #4107    => 'K=5000 L=5000 H=3000 M=10 O=400 E=30 --ambiguous=iupac --notransition --step=20',
   

   • Default parameters of LastZ

   # hsp_threshold (K)      = 3000
   # gapped_threshold (L)   = 3000
   # x_drop (X)             = 910
   # y_drop (Y)             = 9400
   # gap_open_penalty (O)   = 400
   # gap_extend_penalty (E) = 30