Team II Functional Annotation Group

Team 2: Functional Annotation

Team Members: Danielle Temples, Courtney Astore, Rhiya Sharma, Ujani Hazra, Sooyoun Oh

Introduction

What is Functional Annotation?

The practice of putting biological meaning to coding genes (genes that encode proteins) and their corresponding protein sequences. Such annotations can be derived using homology and ab initio based approaches, which will be further explained in subsequent sections.

Objective: Perform a full functional annotation on the genes and proteins determined by the Gene Prediction group that is relevant to C. jejuni

Homology Approaches

• Determine function via sequence similarity to already functionally annotated sequences
• Limited by what we already know.

Ab Initio Approaches

• Determine function via predictive model without comparing to existing sequences
• Based on laws of nature
• Difficult to verify without experiments

Data Overview

We received 50 fna and 50 faa files from the gene prediction group. The 50 fna files are multifasta files representing each genome. The 50 faa files are multifasta files representing each proteome.

Clustering

• Significant sequence similarity implies shared ancestry that often leads to shared function
• Clustering such sequences can reduce repeat queries in homology-based annotations
• Reducing repeats improves speed and storage costs

CD-HIT

• Widely used program for clustering and comparing protein or nucleotide sequences
• Very fast and can handle extremely large databases

./cd-hit -i [input file] -o [output file name]

• We were able to get 90305 protein sequences down to 7414 representative clusters.
• We were able to get 90305 DNA sequences down to 3989 representative clusters.

Homology Methods

Categories

Prophage:

• Play an important role in the evolution of bacterial genomes and their pathogenicity​
• Can change or knock out gene functions; alter gene expression

Virulence:

• A pathogen's ability to infect or damage a host
• Ex: toxins, surface coats that inhibit phagocytosis, surface receptors that bind to host cells

Fully Automated Functional Annotation:​

• Tools that annotate a spectrum of features related to the function

Antibiotic Resistance

• When bacteria develop the ability to defeat the drugs designed to kill them
• Leads to higher medical costs, prolonged hospital stays, and increased mortality​

Operons:

• A functional unit of transcription and genetic regulation​
• Identifying these may enhance our knowledge of gene regulation & function which is a key addition to genome annotation

eggNog-Mapper

• Evolutionary Genealogy of Genes: Non-supervised Orthologous Groups
• Functional Annotation of large sequence sets by fast orthology assignments using precomputed clusters and phylogenies from the eggNOG database
• Use HMM-based search and Diamond based searches to lead to the best seed ortholog in eggNOG
• Orthology predictions for functional annotation results in a higher precision than traditional homology searches by avoiding transferring annotations from close paralogs

Run eggNog-mapper for clustering results against eggNOG bacteria database.

./emapper.py  -i <cluster> --output <result> -d bact -m diamond

VFDB

• Virulence Factor DataBase
• Provide virulence structure features, functions, and mechanisms used to allow pathogens to conquer new niches and circumvent host defense mechanisms
• BLAST based identification of virulence genes

Download VFDB core dataset (includes genes associated with experimentally verified VFs only) and convert into BLAST database.

makeblastdb -in VFDB_db -dbtype 'nucl' -out <db_name>

Run BLASTn for clustering results against VFDB database.

blastn -db <db_name> -query <cluster> -out <result> -max_hsps 1 -max_target_seqs 1 -outfmt 6 -perc_identity 100 -num_threads 5

CARD

• Comprehensive Antibiotic Resistance Database​
• Provides data, models, and algorithms relating to the molecular basis of antimicrobial resistance​
• Can be used for the analysis of genome sequences using the Resistance Gene Identifier

Install the RGI software using conda & download CARD database:

wget https://card.mcmaster.ca/latest/data
tar -xvf data ./card.json

Build a local database:

rgi load --card_json <path to card.json> --local

Run RGI main with protein sequences:

rgi main -i <path to cluster.faa> -o <output_file_name> -t protein --local

Create a tab-delimited file from rgi results:

rgi tab -i <path to output_file_name.json>

MicrobesOnline

Predicts whether pairs of adjacent genes that are on the same strand are in the same operon, based on the intergenic distance between them, whether orthologs of the genes are near each other in other genomes, and their predicted functions.

• Download all the operon tables for Campylobacter jejuni.​
• Use GID to retrieve the protein reference sequences.​
• Create a database using BLAST:

makeblastdb -in <fasta file > -dbtype prot -out <database>

• Query the clustered sequences with the reference protein database:

blastp -query cdhit/faa_rep_seq.faa -db tmp/db_operon -evalue 0.01 -max_target_seqs 1 -max_hsps 1 -outfmt 6 -out tmp/hits_0.01.txt -num_threads 5

Ab Initio Methods

Categories

Transmembrane Proteins (Cell Membrane and Outer Membrane):​

• Bacteria have the ability to export effector proteins in membranes of eukaryotic host
• Integral membrane protein that function as gates or docking sites that allow or prevent the entry or exit of materials across the cell membrane​

Signal Peptides:

• Guide secretory proteins to find their correct locations outside the cell membrane for signal transduction 

CRISPR:​

• Provides immunity to the bacteria against Bacteriophages
• Contributes to the Virulence and Pathogenicity of the bacteria

TBBpred

• Uses neural networks/SVM approach
• Predicts the transmembrane Beta barrel regions in a given protein sequence

TMHMM2

• Uses Hidden Markov Model​ approach
• Outputs the number of transmembrane helices predicted, the position of each residue with respect to the cell membrane (outside/inside the cell, transmembrane segment), and optional graphical output visualizing the predicted helix​

Input a multifasta file into TMHMM.

tmhmm <input multifasta file> > <output file> 

PilerCR

• Rapid identification and classification of CRISPR repeats
• High sensitivity and high specificity

Input a multifasta file into PileCR.

pilercr –in <input multifasta file> -out <output file> -noinfo –quiet

SignalP

• Uses a deep neural network algorithm
• Predicts presence of signal peptides and location of cleavage sites
• Does not determine lipoproteins

Input a fasta file into SignalP.

signalp –fasta <input_sequence_file> -org gram- -format short –gff3

Initial Pipeline

Results

Clustering

Ab Initio

Transmembrane Proteins

• Minimum = 810
• Maximum = 1,049
• Average = 928.8

Signal Peptides

• Minimum = 71
• Maximum = 99
• Average = 94.3

Lipoproteins

• Minimum = 43
• Maximum = 67
• Average = 52.8

CRISPR

• Minimum = 0
• Maximum = 2
• Average = 0.8

Homology

Antibiotic Resistance

• Minimum = 1
• Maximum = 10
• Average = 3.8

Virulence Factors

• Total = 132
• Minimum = 1
• Maximum = 13
• Average = 3.1

Operons

• Minimum = 505
• Maximum = 856
• Average = 678.1

eggNog-mapper Ouput

EggNog-mapper provides useful functional annotations such as seed eggNOG ortholog, seed ortholog evalue/score, predicted taxonomic group, predicted protein name, GO terms, KEGG ko, KEGG Pathways, eggnog OG, COG Functional Category, free text descriptions, and many more. The descriptions of the COG Functional Categories are shown below.

eggNog-mapper Results

• Total = 44,098
• Minimum = 669
• Maximum = 1087
• Average = 881.9

Final Pipeline