TOGA is a method that integrates gene annotation, inferring orthologs, and classifying genes as intact or lost. We want to use TOGA to detect genetic innovations in Lasallia pustulata. Therefore, we are searching for genes that are present in Lasallia hispanica but lost in Lasallia pustulata.

1. Create an own environment for TOGA and install nextflow:

   conda create --name toga python=3.7.3
   conda activate toga  
   conda install -c bioconda nextflow 
   conda install -c anaconda biopython 

2. Clone TOGA from github and install it:

   git clone https://github.com/hillerlab/TOGA.git
   cd TOGA
   python3 -m pip install -r requirements.txt --user
   ./configure.sh
   # ignore the warnings
   ./run_test.sh micro

   Your output should look something like this in the end:

   #### STEP 11: Cleanup: merge parallel steps output files
   Saved results to /home/ecoevo01/TOGA/TOGA/micro_test_out
   Done! Estimated time: 0:01:40.027813
   Program finished with exit code 0
   
   JH567521	463144	506100	ENST00000262455.1169	1000	-	463144	506100	0,200,255	8	102,103,142,112,117,58,116,185,0,1982,30295,31351,36911,38566,41322,42771,
   JH567521	395878	449234	ENST00000259400.1169	1000	+	395878	449234	0,0,200	7	123,66,226,116,51,87,240,	0,11871,38544,45802,45994,52305,53116,
   JH567521	299723	336583	ENST00000618101.1169	1000	+	299723	336583	0,0,200	7	28,923,130,173,200,179,248,	0,1256,6085,6677,19146,21311,36612,
   Success!

TOGA uses at least 3 different input data:

1. Gene annotation of the reference genome, in our case from L.hispanica, as a bed-12 file
2. Genome alignment between the reference (L.hispanica) and query genome (L.pustulata) as a chain file
3. Reference and query genome sequences as 2bit files

We need to create the required inputs for TOGA from our assemblies. This require different tools, including lastz, axtChain, chainNet, chainSort, chainCleaner, faToTwoBit, gff3ToGenePred, GenePredToBed, RepeatFiller and some self-written scripts. The tools from UCSC (such as faToTwoBit, axtChain,..) are pre-compiled and can be downloaded here using the command below. Please note that you need to find the correct name for the tools that are displayed on that website, for example lastz-1.04.00 for lastz.

rsync://hgdownload.soe.ucsc.edu/genome/admin/exe/linux.x86_64/<tool_name> ./ 

We have downloaded and installed all of them for you Add the following folder to you variable $PATH by adding this line to your ~/.bashrc file:

export PATH=/share/gluster/Projects/vinh/ecoevo/src/for_toga:$PATH

Then, apply the change with source ~/.bashrc (or restart the terminal) and re-activate toga conda environment.

Now, we can generate the input files for TOGA.

1. First of all, we convert our assemblies from L. pustulata and L. hispanica into 2bit files using faToTwoBit:

   faToTwoBit L_hispanica_assembly.fasta L_hispanica.2bit
   faToTwoBit L_pustulata_assembly.fasta L_pustulata.2bit

2. As the next step, we do the whole-genome alignment with the workflow mentioned in Pippel et. al. 2020 that requires different steps and tools
   1. Do genome alignment with lastz. This will take about 6 hours, please use Slurm with 1 CPU and 4GB Ram. While lastz is running, you can do step 2c and 3

      lastz L_hispanica.2bit L_pustulata.2bit --action:target=multiple --hspthresh=2400 --gappedthresh=3000 --format=axt --out=hispanica_pustulata_aln.axt 

   2. Compute chain file from axt alignments using axtChain

      axtChain hispanica_pustulata_aln.axt L_hispanica.2bit L_pustulata.2bit hispanica_pustulata_aln.chain -linearGap=medium

   3. Get length of each sequence in the L_hispanica and L_pustulata assembly using BASH or the scripts below. You will need to:

      python /share/gluster/Projects/vinh/ecoevo/src/for_toga/fasta_len.py <LasHis.fa> <LasHis.sizes>
      python /share/gluster/Projects/vinh/ecoevo/src/for_toga/fasta_len.py <LasPus.fa> <LasPus.sizes>
      # make sure that the identifiers in your .sizes files match the ones in your .axt file

   4. Compile chain nets with chainNet

      # run chainSort
      chainSort hispanica_pustulata_aln.chain hispanica_pustulata_aln.sorted.chain
      # run chainNet
      chainNet hispanica_pustulata_aln.sorted.chain L_hispanica.sizes L_pustulata.sizes hispanica.net pustulata.net

   5. Improve genome alignment using chainCleaner

      chainCleaner hispanica_pustulata_aln.chain L_hispanica.2bit L_pustulata.2bit hispanica_pustulata_aln.cleaned.chain hispanica_pustulata_aln.bed -net=hispanica.net linearGap=medium

   6. Comprehensively align repetitive genomic regions with RepeatFiller (using Slurm again with 1 CPU and 4GB memory. This will take about 1.5 hours)

      RepeatFiller.py -c hispanica_pustulata_aln.cleaned.chain -T2 L_hispanica.2bit -Q2 L_pustulata.2bit | tee hispanica_pustulata_aln.cleaned.filled.chain

   7. The output from RepeatFiller (saved in hispanica_pustulata_aln.cleaned.filled.chain) is the chain file that we want to use as an input for TOGA. Sometimes the RepeatFiller output includes empty lines which must be deleted. Use the following command:

      sed -i '/^$/d' hispanica_pustulata_aln.cleaned.filled.chain

3. Convert the annotation gff3 file of the reference genome L. hispanica into bed file using gff3ToGenePred and GenePredToBed

   gff3ToGenePred L_hispanica.gff3 L_hispanica.genePred
   genePredToBed L_hispanica.genePred L_hispanica.bed

4. After those steps, you will have a lot of files. We suggest that you copy these 4 input files for TOGA to an additional folder before running the tool:
   1. The annotation in bed format L_hispanica.bed
   2. The genome alignment as chain file hispanica_pustulata_aln.cleaned.filled.chain
   3. 2 genome sequences in 2bit format L_hispanica.2bit and L_pustulata.2bit

Use a Slurm script with 12 CPUs and 8GB memory to run TOGA

python3 /path/to/your/downloaded/TOGA/toga.py hispanica_pustulata_aln.cleaned.filled.chain L_hispanica.bed L_hispanica.2bit L_pustulata.2bit

TOGA produces different output files that you can look up here. The most interesting output file for us is loss_summ_data.tsv, which contains the classifications for each projection, transcript, and gene. TOGA identifies the following classes:

    N - no data due to technical reasons (such as CESAR memory requirements)
    PG - no orthologous chains identified, TOGA projected transcripts via paralogous chains and cannot make any conclusion
    PM - partial & missing. Most of the projection lies outside scaffold borders
    L - clearly lost (several inactivating mutations)
    M - missing, assembly gaps mask >50% of the prediction CDS
    G - "grey", there are inactivating mutations but not enough evidence for "clearly lost" class. In other words: neither lost nor intact
    PI - partially intact: some fraction of CDS is missing, but most likely this is intact
    I - clearly intact
    UL - uncertain loss (few inactivating mutations localized in a single exon)

We are interested in the genes that are present in L. hispanica but lost in L. pustulata not by technical artifacts (e.g. errors from gene annotation process).

1. Search for TRANSCRIPTs and identify all gene IDs that are marked as L (lost), M (missing), PM (partial missing) or UL (uncertain loss) in the file loss_summ_data.tsv. These four sets of genes will form our genes of interest
2. Use genome viewer of g-nom to investigate some of those genes and their gene neighborhood in the L. hispanica contig assembly

From our previous analysis, we obtained a list of genes, which are missing in L. pustulata but present in L. hispanica. But,

1. What are they, or, what are their characteristics?
2. How old are they, or do they loss only in L. pustulata or also in other taxonomic clades? In other words, are they L. pustulata specific losses or L. hispanica specific gains?

In order to answer those questions, we are going to study their phylogenetic profiles across a wide range of species by using fDOG and PhyloProfile. fDOG is a directed ortholog search tool, which can look for orthologs of a single gene of interest in a set of search taxa. These search organisms can be diverse, from closely related to distantly taxa to our reference species, from which the seed genes are originated.

The taxon set we will use for this analysis consists of L. hispanica, L. hispanica, the close related species U. muehlenbergii, and the set of 78 QfO taxa (what is the “QfO”? Why do we work with these 78 taxa?).