Identification of viruses and other pangenomic elements within SAGs

In this tutorial we are focusing on viruses from plates AG-910.

Notebooks 1, 2 and 3 will take you through virus searching for and curation of viruses within an indivial SAG.

To get started, open a terminal and navigate into your working directory:

cd ~/storage/userlab/{your name here}

then clone this github repository:

git clone https://github.com/Bigelow-eSCG-tutorials/day3_virus_tutorial.git

and navigate into this directory in terminal:

cd day3_virus_tutorial

Navigate into this directory in the jupyterhub navigation pane on the left side of your screen, select 1_finding_phages.ipynb to open it in a new window in jupyterhub and let's find some viruses.

---

Virus Finding

VirSorter2:

source activate /mnt/storage/envs/vs2
virsorter run -i ${contigs_fasta} --include-groups "dsDNAphage,ssDNA"

DeepVirFinder:

source activate /mnt/storage/envs/dvf
python /mnt/storage/envs/opt/DeepVirFinder/dvf.py -i ${contig} -o ${plateout} -c ${cpus}

---

Identification of initial candidates

Results from the above workflows were merged, contigs were maintained as initial viral candidates based on the following criteria:

• VirSorter2: max_score = 0.9, hallmark genes > 0
• DeepVirFinder: dvf_score_min = 0.9, dvf_pvalue_max = 0.05

---

Candidate initial QC

Initial virus candidate contigs were then extracted and written to a new fasta file and examined by CheckV:

CheckV

source activate /mnt/storage/envs/checkv
checkv end_to_end $infa $outdir -t 20 -d /mnt/storage/reference_dbs/checkv/checkv-db-v1.5

---

Annotation

DRAM

source activate /mnt/storage/envs/dram
DRAM-v.py annotate -i ${contig} -o $plateout --min_contig_size 2000

---

Summary Tables

sag_vcontig_summary.csv

Columns:
sag: sag ID
total_vcontig_bp: total length of viral candidates sequences in SAG
'Complete', 'High-quality', 'Low-quality', 'Medium-quality', 'Not-det ermined’: contig count for checkV quality categories
'Total vContigs’: total number of viral candidate contigs in SAG
'dramv_top_vir_genome_match’: viral genome from NCBI with the most hits to vCandidate open reading frames (orfs)
'dramv_top_vir_pct_matched’: percent orfs matching the top hit virus
'total_vc_orfs’: total number of orfs on virus candidate contigs
'total_annotated_vc_orfs’: total vCandidate orfs that were annotated by DRAMv
'pct_annotated’: percent annotated orfs on virus candidates
'phage structurals’: list of identified viral structural genes on vCandidates
'plate_id’: SAG plate ID
'integrase_count’: total number of integrases on vCandidates within SAG

vcandidate_contig_summary_stats.csv

Columns:
'contig’: contig ID
'sag’: SAG id
'plate’: plate id
from virsorter2
'type_vs2’: virsorter2 type (integrated or individual)
'vs2_pos’: 1 if virsorter identified
from vibrant
'type_vib’: VIBRANT type (X or Y)
'vib_pos’: 1 if VIBRANT identified
from deepvirfinder. 'dvf_pos’: 1 if DeepVirFinder identified
added columns
'consensus_score’: count of how many virus finders identified as viral (max 3)
'contig_length’: contig length in bp
from checkV. 'provirus’: 'Yes' if checkv identified as integrated
'gene_count’: total genes on contig
'viral_genes’: total viral genes on contig (checkV determined)
'host_genes’: total host genes on contig (checkV determined)
'pct_host_genes’: percent genes on contig identified as ‘host’ by checkV
'checkv_quality’: checkV determined viral genome quality
'miuvig_quality’: another genome quality metric
'completeness': checkv estimated viral contig completeness
'completeness_method': method checkv used to estimate completeness

per-SAG ORF annotation tables

'orfid': DRAM-v open reading frame ID
'fasta': Originating fasta file
'scaffold': Scaffold/contig ID
'gene_position': Gene position on contig
'start_position': Gene start position on contig
'end_position': Gene end position
'strandedness': gene strandedness
'rank': strength of the metabolic annotation -- see (here)[https://github.com/WrightonLabCSU/DRAM/wiki/4a.-Interpreting-the-Results-of-DRAM#the-annotations-master-table]
Blast style database searches:
'viral_id'
'viral_hit'
'viral_RBH'
'viral_identity'
'viral_bitScore'
'viral_eVal'
'peptidase_id'
'peptidase_family'
'peptidase_hit'
'peptidase_RBH'
'peptidase_identity'
'peptidase_bitScore'
'peptidase_eVal'
hmm style searches: 'kegg_id'
'kegg_hit'
'pfam_hits'
'cazy_hits'
'vogdb_description'
'vogdb_categories'
'heme_regulatory_motif_count'
'vogdb_hits'
'is_transposon'
'amg_flags'
added columns
'sag': sag id
'vogdb_text': extracted vogdb text description
'vog_db_vir_protein': identified as a viral protein by columns 'vogdb_categories'
'viral_hit_genome': extracted genome description for the viral hit
'viral_hit_gene': extracted gene text for the viral hit
'viral_hit_gene_desc': extracted gene description for the viral hit
'is_integrase': identified as integrase in any annotation column (1 if integrase, 0 if not)
'annotation': summarized gene annotation from all database comparison results
'annotation_source': column that annotation was extracted from