Fusarium Lifestyles

Bioinformatics analysis pipeline for:

Hill et al. (2022) Lifestyle transitions in fusarioid fungi are frequent and lack clear genomic signatures. Molecular Biology and Evolution 39(4):msac085. doi:10.1093/molbev/msac085.

The pipeline was written for and run on Queen Mary University of London's Apocrita HPC facility which uses the Univa Grid Engine batch-queue system. This means that many of the bash scripts (.sh file endings) specify core allocation, run times and memory usage allocation that may need to be adapted for different platforms.

Associated data files:

---

1 Assembly

cd assembly

1.1 Read quality control

cd assembly/reads

Requires raw fastq.gz paired-end reads in this directory as well as TruSeq3-PE.fa file with adapter sequences downloaded from here (for Illumina NovaSeq 6000 151bp paired-end reads).

1. qsub trimmomatic.sh - trims raw reads using Trimmomatic.
2. qsub fastqc.sh - after trimming, checks read quality with FastQC.

1.2 De novo genome assembly

cd assembly/denovo_assembly

1. ./submit_assembly.sh - makes new directory and submits job scripts for each assembly tool - abyss.sh (ABySS), megahit.sh (MEGAHIT) and spades.sh (SPAdes).
2. ./abyss_comp.sh - compares the assembly stats to choose 'best' kmer size for ABySS (must be done after abyss.sh has finished for all kmer sizes and strains).

1.3 Polishing

cd assembly/polishing

1.qsub polish.sh - for each strain and assembly tool, maps raw reads to assembly and calculates mapping statistics with BWA-MEM and SAMtools and polishes the assembly with Pilon. Also removes sequences <200bp using Seqtk for NCBI compliance.

After completing 4 Assessment and uploading to NCBI:

2../ncbi_filter.sh - removes sequences identified as mitochondrial or duplicates by NCBI (listed in files saved as duplicates_* and mito_remove_* for each strain) and trims sequences identified as having mitochondrial contaminants (listed in files saved as mito_trim_*.bed) using bedtools.

1.4 Assessment

cd assembly/assessment

1. ./submit_assessment.sh - submits quast.sh (QUAST), busco.sh (BUSCO) and blast.sh (BLAST) scripts for assembly quality statistics. busco.sh requires the Hypocreales BUSCO dataset downloaded from here).
2. qsub blobtools.sh - submits blobtools.sh to run BlobTools (must be done after blast.sh has finished for all strains).

---

2 Annotation

cd annotation

2.1 Repeatmasking

cd annotation/repeat_masking

1. qsub repeatmodeler.sh - makes custom repeat library for each strain using RepeatModeler.
2. qsub repeatmasker.sh - uses the custom repeat libraries to softmask assemblies using RepeatMasker.

2.2 MAKER pipeline

cd annotation/maker

Informed by this tutorial. Requires ESTs and proteins from Fusoxy1 and Fuseq1 (Mesny et al. 2021) downloaded from Mycocosm in this directory.

Round 1

cd annotation/maker/round1

qsub maker.sh - submits first run of MAKER using ESTs and proteins, as indicated in .ctl files.

Round 2

cd annotation/maker/round2

1. qsub training_snap/snap.sh - trains SNAP using gene models from the first MAKER round.
2. qsub maker2.sh - submits second run of MAKER using trained SNAP (as indicated in .ctl files).

Round 3

cd annotation/maker/round3

1. qsub training_snap2/snap2.sh - trains SNAP again using gene models from the second MAKER round.
2. qsub maker3.sh - submits third run of MAKER using second trained SNAP (as indicated in .ctl files).
3. qsub rename.sh - after obtaining unique locus tags from e.g. NCBI (see locus_tags.txt), renames IDs in gff and fasta files.
4. qsub gag.sh - runs GAG to remove introns <10bp, remove terminal Ns and correct start and stop codons in gff file for NCBI compliance.

---

3 Orthology inference

cd orthology_inference

1. ./submit_protein_download.sh - submits ncbi_ftp_links.r and protein_download.sh to download of predicted protein sets of Fusarium strains from NCBI.
2. qsub orthofinder.sh - submits orthology inference using OrthoFinder.

---

4 Phylogenomics

cd phylogenomics

1. ./submit_alignment.sh - submits alignment.sh for alignment of single copy orthogroups from OrthoFinder with MAFFT followed by trimming with BMGE and trimAl.
2. ./concat.sh - concatenate single copy orthogroup alignments and prepare partition files using AMAS.
3. ./submit_modeltestng.sh - submits modeltest-ng/modeltestng.sh to run ModelTest-NG on all single copy orthogroups (in computationally tractable chunks).
4. ./submit_speciestrees_concatenation.sh - submits concatenation-based species tree methods - species_tree/raxmlng.sh (RAxML-NG) and species_tree/iqtree.sh (IQ-TREE).
5. ./submit_RAxML-NG_genetrees.sh - submits RAxMLNG_genetrees.sh to run RAxML-NG for individual gene trees.
6. ./submit_speciestrees_coalescent.sh - submits coalescent-based species tree methods - species_tree/astral.sh (ASTRAL-III) and species_tree/astral-pro.sh (ASTRAL-Pro) using genes trees.

---

5 Divergence time estimation

cd divergence_time_estimation

5.1 SortaDate

cd divergence_time_estimation/sortadata

qsub sortadate - reroots gene and RAxML-NG species tree and runs with SortaDate to filter for top ten 'clock-like' genes.

5.2 MCMCTree

cd divergence_time_estimation/mcmctree

1. qsub mcmctree_dating_step1.sh - adds secondary time calibrations to species tree nodes and submits first step of approximate likelihood divergence time estimation with protein data using MCMCTree from PAML (see tutorial).
2. Rscript estimate_rate.r - estimates the scaling parameter for the substitution rate prior to be added to mcmctree_step2_independent.ctl and mcmctree_step2_correlated.ctl.
3. ./submit_mcmctree_dating_step2.sh - submits mcmctree_independent.sh and mcmctree_correlated.sh for second step of approximate likelihood estimation for both independent and correlated rates relaxed clock models.

---

6 CSEP & CAZyme prediction

cd CSEP_CAZyme_prediction

1. ./submit_CSEPprediction.sh - submits all programmes in the CSEP prediction pipeline - signalp/signalp.sh (SignalP), targetp/targetp.sh (TargetP), phobius/phobius.sh (Phobius), tmhmm/tmhmm.sh (TMHMM), prosite/ps_scan.sh (ps_scan), nucpred/nucpred.sh (NucPred), predgpi/predgpi.sh which in turn submits predgpi/PredGPI.r to use the R package ragp (PredGPI) and effectorp/effectorp.sh (EffectorP).
2. ./submit_CSEPfilter.sh - submits CSEPfilter to produce lists of CSEPs from all programme results.
3. ./submit_CSEPblast.sh - submits blastp/blastp.sh to BLAST of CSEPs against the PHI-base database (requires phi-base_current.csv and phi-base_current.fas to be downloaded from here into the blastp directory).
4. ./submit_CAZymeprediction.sh - submits run_dbcan/run_dbcan.sh to run run_dbcan.
5. qsub submit_orthogroupparsing.sh - submits orthogroup_parser.r to make abundance matrices of orthogroups for all strains and categorises whether they are CSEPs/CAZymes and core/accessory/specific.

---

7 Lifestyle comparison

cd lifestyle_comparison

./submit_lifestyletest.sh - submits lifestyle_v_phylogeny.r to prepare input file for lifestyle-test.sh which runs PERMANOVA-based lifestyle test on orthogroup and CSEP presence absence matrices; run_edited.py is modified from the original script run.py by Mesny & Vannier.

---

8 Selection

cd selection

8.1 dN/dS analysis

1. qsub gbff_files/ncbi_gbff_download.sh - downloads GBFF files for the strains used in this study from NCBI; also need Ilysp1 transcripts downloaded from Mycocosm in gbff_files directory.
2. ./submit_pal2nal.sh - submits pal2nal.sh script to pull corresponding nucleotides for all proteins using pull_nucleotides.py and prepares codon alignments using PAL2NAL.
3. ./submit_hyphy.sh - prepares file inputs and submits scripts for HyPhy dN/dS methods - hyphy/BUSTED.sh, hyphy/aBSREL.sh and hyphy/Contrast-FEL.sh.

8.2 Codon optimisation

cd selection/codon_optimisation

1. ./pull_ribosomes.sh - extracts ribosomal protein encoding genes from Fusgr1 and submits blast.sh to run BLAST search against all strains in this study.
2. ./submit_codon_optimisation.sh - submits codon_optimisation.r script to estimate various codon usage bias statistics and codon optimisation values.

---

9 Statistics and data visualisation

Rscript stats_and_plots.r

---

Citation

Hill et al. (2022) Lifestyle transitions in fusarioid fungi are frequent and lack clear genomic signatures. Molecular Biology and Evolution 39(4):msac085. doi:10.1093/molbev/msac085.