The Microbial Comparative Genomics Workflow

The Microbial Comparative Genomics Workflow is composed of two Python scripts: get_genome_data.py and genome_analysis.py.

• The get_genome_data.py script integrates with the Pathosystems Resource Integration Center (PATRIC) to allows users to perform genome assembly and gene annotation from raw sequencing reads or pre-assembled contigs. Also, get_genome_data.py allows users to download bacterial genomes publically available in PATRIC database. The get_genome_data.py script enables multiple genomes to be processed and/or downloaded from PATRIC simultaneously, thereby promoting high-throughput acquisition of genomic data for downstream comparative analyses.

• The genome_analysis.py script allows users to perform several comparative analyses on a group of genomes acquired using get_genome_data.py. The analyses performed by genome_analysis.py are aimed to address how similar or different individual or subgroups of genomes are compared to the larger group. Importantly, genome_analysis.py determines phylogenetic relationship based on the pan-genome and identifies which genes are conserved or unique genes within a genome group. The genome_analysis.py script also integrates DIAMOND BLAST to enable the detection of genes of interest, such as virulence factor or antibiotic resistance genes, to promote investigation into the functional significance of genes.

The following dependencies are required by the Microbial Comparative Genomics Workflow:

• Python 2.7
• PATRIC account
• PATRIC Command Line Interface
• Numpy
• Pandas
• Matplotlib
• Seaborn
• DIAMOND BLAST

get_genome_data.py

The get_genome_data.py script allows users to process novel bacterial genomes and efficiently compile a group of genomes for their downstream comparative analyses by:

1. performing the assembly and annotation of multiple bacterial genomes from raw sequencing reads and/or pre-assembled contigs and
2. downloading genomic data publically available in the PATRIC database.

The get_genome_data.py script uses the Comprehensive Genome Analysis Service hosted by PATRIC in order to automate the assembly and annotation workflow, thereby increasing throughput when numerous genomes must be processed. The get_genome_data.py accepts pair-end reads (fastq) or pre-assembled contigs (fasta), which provides flexibility for the user. The output of get_genome_data.py is a full-genome report summarizing the genome assembly and annotation characteristics, assembled genome sequences (contigs fasta files), gene annotation sequences (DNA and protein fasta files), and annotation metadata. Note: After logging into your PATRIC account online, create new folder called " AssemblyJob" in your home directory (/user@patricbrc.org/home/AssemblyJob). The pair-end reads (fastq) and pre-assembled contigs (fasta) will be uploaded to this directory, and the assembly/annotation data be stored in this directory as well.

To compliment datasets, get_genome_data.py allows user to access and download additional genomic data from the PATRIC database, such as reference genomes. The get_genome_data.py allows researchers to easily acquire assembled genome sequences (contigs fasta files), gene annotation sequences (DNA and protein fasta files), and annotation metadata from numerous genomes quickly by providing the PATRIC genome ids in the metadata table.

Below are the flags available for get_genome_data.py.

Flag	Flag (verbose)	Description
-h	--help	Show all flags and descriptions
-u	--username	Provide username for PATRIC account. Prompt to enter password will appear.
-m	--metadata_file	Specify metadata file.
-f	--file_upload	Upload read and/or contig files? Enter "yes" or "no". Default is "yes". If file with same name has already been uploaded to PATRIC, it will be overwritten by the newly uploaded file.
-a	--assembly_annotate	Execute assembly and annotate pipeline? Enter "yes" or "no". Default is "yes".
-c	--check_job	Check status of assembly/annotation job? Enter "yes" or "no". Default is "yes". When job is complete, genome reports, contigs, and annotations data will be downloaded to output folder.
-d	--download_data	Download genome reports, contigs, and annotations data for assembled/annotated genomes for previously completed jobs to output folder? Enter "yes" or "no". Default is "no".
-p	--patric_download	Download genome reports, contigs, and annotations data for PATRIC genomes? Enter "yes" or "no". Default is "yes".
-o	--output_folder	Specify output folder for downloaded data.

Below is an outline of the assembly and annotation workflow for get_genome_data.py.

genome_analysis.py

The genome_analysis.py script is used to determine the phylogenetic relationship of genomes and identify which genes are core or unique genes in the pan-genome. As a result, genome_analysis.py allows users to investigate the similarities and difference of individual or subgroups of genomes with the larger group. The genome_analysis.py script also integrates DIAMOND BLAST to enable the detection of genes of interest, such as virulence factor, antibiotic resistance or custom gene databases, that will assist users when interpreting the output data. The genome_analysis.py uses global protein families (i.e., PGfams) determined by PATRIC as the basis for determining relationships in the pan-genome.

Below is an outline of the workflow for genome_analysis.py.

Below is summary of the analyses performed by genome_analysis.py.

1. Pan-genome phylogenetic tree

Creates PHYLIP formatted matrix of binary presence and absence of PGfams in the pan-genome of the genome group. The output file is *pan-genome-tree_out.txt * and can be used to make a phylogenetic tree in other programs such as RAxML, IQ-TREE, or similar program.

2. Pan-genome hierarchically-clustered heatmap

A hierarchically-clustered heatmap (i.e., clustermap) of PGfams in the pan-genome is created using the Seaborn clustermap function. Cluster analysis is performed using the “Euclidean” distance metric and the “average” linkage method on both axis (i.e., genomes and PGfams). The output is an image of the clustermap (i.e., clustermap.png), which shows the abundance of the genes per genome and their clustering relationships. (Please note that the depiction of the dendrogram for the PGfam relationships is excluded in clustermap.png image because its dendrogram is too busy.) The DataFrame corresponding to the clustermap is generated (i.e., PGfam_clustermap_dataframe.txt), which can be analyzed if desired.

3. Core and unique genes in pan-genome

The core (i.e., present in all genomes) and unique (i.e., present in only a single genomes) PGfams in the pan-genome are determined and merged into the annotation metadata per genome.

4. Unique genes in genome subgroups

PGfams unique to subgroups of genomes within the larger group are identified and merged into the annotation metadata for each genome. The subgroups are indicated in the metadata table. Sometimes subgroups of interest may be known a priori or may be realized following analysis of the pan-genome phylogenetic tree* and/or pan-genome hierarchically-clustered heatmap results.

5. Gene copies versus median gene copy number in genome group

The median copy number for each PGfams in the genome group is calculated and then compared per genome. The resulting analysis yields how many PGfams are present in a genome and if the number PGfams is equal to, greater than, or equal to the median. Results are merged into the annotation metadata for each genome

6. Diamond BLAST for virulence factors, antibiotic resistance, and other genes

DIAMOND BLAST is used to determine if genes are homologous to known virulence factor or antibiotic resistance genes. Virulence factor genes are derived from the Virulence Factor DataBase (VFDB), and antibiotic resistance genes are derived from Comprehensive Antibiotic Resistance Database (CARD). Users can also search against a custom database by providing a multiple sequence protein fasta file. The best match hits per gene are merged into the annotation metadata for each genome.

For each genome, the above analyses are also merged into a single annotation metadata file, thereby allowing all results to be viewed simultaneously.

Below are the flags available for genome_analysis.py.

Flag	Flag (verbose)	Description
-h	--help	Show all flags and descriptions
-i	--input_folder	Specify folder with annotation data.
-o	--output_folder	Specify name for output folder.
-m	--metadata_file	Specify metadata file.
-make_tree	--make_tree	Generate binary matrix of protein families in PHYLIP format for pan-genome phylogenetic tree? Enter "yes" or "no". Default is "yes".
-make_clustermap	--make_clustermap	Create hierarchically-clustered heatmap (ie clustermap) of protein families in pan-genome? Enter "yes" or "no". Default is "yes".
-core_unique_genes	--core_unique_genes	Determine core and unique protein family genes for genome group? Enter "yes" or "no". Default is "yes".
-median_analysis	--median_analysis	Calculate median protein family gene copy number in genome group and if protein family genes for individual genomes are equal to, greater than, or less than median? Enter "yes" or "no". Default is "yes".
-subgroup_genes	--subgroup_genes	Determine unique protein family genes for genome subgroup within the larger genome group? Enter "yes" or "no". Default is "yes".
-VF_blast	--VF_blast	Perform DIAMOND blast analysis for virulence factor genes? Enter "yes" or "no". Default is "yes".
-res_blast	--res_blast	Perform DIAMOND blast analysis for antibiotic resistence genes? Enter "yes" or "no". Default is "yes".
-custom_blast	--custom_blast	Perform DIAMOND blast analysis for custom gene database? Enter "yes" or "no". Default is "no".
-custom_fasta	--custom_fasta	Provide custom gene database as multi-sequence fasta file using amino acids.
-merge_all_annotations	--merge_all_annotations	Merge all annotation metadata files in output folder? Enter "yes" or "no". Default is "yes".

Example/Tutorial

In this section, an example using The Microbial Comparative Genomics Workflow is described. All input data and output results are available in the repository here.

E. coli commensal intestinal bacterial found in many species including humans and other mammalian species. Some E. coli strains encode virulence factor genes that allow make them pathogenic. One virulence factor gene expressed by some E. coli strains is colibactin, a genotoxin produced by pks gene island, and these strains are associated with UTI, meningitis, and colon cancer in humans and animal models. In this example, the get_genome_data.py and genome_analysis.py scripts are used to compare genomes of human and rodent E. coli strains that do and do not encode the pks gene island.

More specifically, we would like to know:

• Are E. coli strains from human and rodents phylogenetically distinct?
• What genes are found in all pks+ E. coli genomes?
• What genes are found only in genomes of E. coli strains isolated from all rodents?
• What genes are found only in novel genomes of E. coli strains isolated from the rodent?
• What genes are unique to novel genome of E. coli strains isolated from a rat compared to mice?

Here, get_genome_data.py will be used to assembly and annotate six user genomes. Pre-assembled contigs from five E. coli strains isolated from lab mice (Mannion A. et al. Genome Announc. 2016.) and Illumina MiSeq 2x250 bp reads from one E. coli strain isolated from pet rat (Fabian NJ. et al. Vet Microbiol. 2020.). The sequencing reads (fastq files) and pre-assembled contigs (fasta files) along with their corresponding genome names are recorded in the metadata table shown below. All of these isolates, except one, were experimentally shown to encode the pks gene island.

To complement our genome group, data from four reference E. coli genomes will be acquired from PATRIC using get_genome_data.py. These genomes will include:

• E. coli K12: non-pathogenic lab strain
• E. coli NC101: pks+ strain isolated from mouse and associated with colon cancer
• E. coli IHE3034: pks+ strain from human meningitis
• E. coli UTI89: pks+ strain from human urinary tract infection

PATRIC is browsed to find these four genomes, and their corresponding genome names and genome ids are recorded in the metadata table shown below.

Since we are curious about the similarities and differences between human versus rodent E. coli strains as well as genomes that do and do not encode the pks gene island, we use genome_analysis.py to identify genes that are unique to these subgroups compared to the larger genome group. The subgroups of interested are 1) all rodent genomes, 2) the novel rodent genomes, and 3) all pks+ genomes. These subgroups are recorded in the metadata table shown below.

metadata_table.txt file

genome_name	genome_ids_patric	genome_name_patric	contigs	genome_name_contigs	R1	R2	genome_name_reads	subgroup	include_all_rodents	exclude_all_rodents	include_novel_rodents	exclude_novel_rodents	include_all_pks	exclude_all_pks
Escherichia coli IHE3034	714962.3	Escherichia coli IHE3034						all_rodents		Escherichia coli IHE3034		Escherichia coli IHE3034	Escherichia coli IHE3034
Escherichia coli NC101	753642.3	Escherichia coli NC101						novel_rodents	Escherichia coli NC101			Escherichia coli NC101	Escherichia coli NC101
Escherichia coli str. K-12 substr. MG1655	511145.12	Escherichia coli str. K-12 substr. MG1655						all_pks		Escherichia coli str. K-12 substr. MG1655		Escherichia coli str. K-12 substr. MG1655		Escherichia coli str. K-12 substr. MG1655
Escherichia coli UTI89	364106.8	Escherichia coli UTI89								Escherichia coli UTI89		Escherichia coli UTI89	Escherichia coli UTI89
Escherichia coli strain 1408270010			Escherichia_coli_strain_1408270010_contigs.fasta	Escherichia coli strain 1408270010					Escherichia coli strain 1408270010		Escherichia coli strain 1408270010		Escherichia coli strain 1408270010
Escherichia coli strain 1409150006			Escherichia_coli_strain_1409150006_contigs.fasta	Escherichia coli strain 1409150006					Escherichia coli strain 1409150006		Escherichia coli strain 1409150006		Escherichia coli strain 1409150006
Escherichia coli strain 1409160003			Escherichia_coli_strain_1409160003_contigs.fasta	Escherichia coli strain 1409160003					Escherichia coli strain 1409160003		Escherichia coli strain 1409160003			Escherichia coli strain 1409160003
Escherichia coli strain 1512290008			Escherichia_coli_strain_1512290008_contigs.fasta	Escherichia coli strain 1512290008					Escherichia coli strain 1512290008		Escherichia coli strain 1512290008		Escherichia coli strain 1512290008
Escherichia coli strain 1512290026			Escherichia_coli_strain_1512290026_contigs.fasta	Escherichia coli strain 1512290026					Escherichia coli strain 1512290026		Escherichia coli strain 1512290026		Escherichia coli strain 1512290026
Escherichia coli strain 20170221001					20170221001_R1.fastq	20170221001_R2.fastq	Escherichia coli strain 20170221001		Escherichia coli strain 20170221001		Escherichia coli strain 20170221001		Escherichia coli strain 20170221001

First, we acquire our genome data by get_genome_data.py to assemble/annotate our rodent genomes and download our reference genomes from PATRIC. To do so, the following command is used in the PATRIC Command Line Interface:

python get_genome_data.py –u user@email.org –m metadata_table.txt –o Example_data

(Note: The metadata table should be in the working directory. The input reads (fastq files) or contigs (fasta files) can be in the working directory or in another directory as long as the path is indicated in the metadata table. The output folder will be produced in the working directory.)

The user will prompted to enter their PATRIC account password. Genomes from PATRIC will be then downloaded first. Afterwards, reads (fastq files) and contigs (fasta files) will be uploaded to PATRIC. (The Assembly/annotation jobs will then be sent to PATRIC. The status of the jobs will be checked every five minutes. Once a job has completed, the data will be downloaded to the output folder (i.e., ‘”Example_data”).

Sometimes jobs may take several hours to complete. The command terminal can be left open until all jobs are complete and the data has been downloaded. Alternatively, the script can be terminated and the resumed later. Jobs will continue to be process by the PATRIC server. The get_genome_data.py can be run again using different flags to bypass steps already completed in the workflow and begin checking job status again or directly download the data if all the jobs are complete.

The following command can be used in the PATRIC Command Line Interface to begin checking job status again and download data when jobs are completed:

python get_genome_data.py –u user@email.org –m metadata_table.txt –o Example_data -f no -a no –p no

The following command can be used in the PATRIC Command Line Interface to download data for jobs that are completed without checking job status:

python get_genome_data.py –u user@email.org –m metadata_table.txt –o Example_data -f no –a no –c no –d yes –p no

The genomes processed for the assembly/annotation each have five output files:

1. Full genome report (FullGenomeReport.html), which summarizes genome assembly and gene annotation results
2. Assembled genome sequence (contigs.fasta file)
3. Gene annotations in nucleotides sequence (DNA.fasta)
4. Gene annotations in protein sequence (protein.fasta)
5. Annotation metadata (annotation.txt)

The genomes download from PATRIC have four output files:

1. Assembled genome sequence (contigs.fasta file)
2. Gene annotations in nucleotides sequence (DNA.fasta)
3. Gene annotations in protein sequence (protein.fasta)
4. Annotation metadata (annotation.txt)

Now that have the data for all our genomes of interest, we will use genome_analysis.py to perform the comparative analyses. The following command is used:

python genome_analysis.py –i Example_data –o Example_analysis –m metadata_table.txt

(Note: The metadata table should be in the working directory. The output folder will be produced in the working directory. The The DIAMOND BLAST program and databases (dmnd file or multi-sequence fasta file for a custom database) should be in working directory.)

The output for each analysis in genome_analysis.py are the following:

1. Pan-genome phylogenetic tree: pan-genome-tree_out.txt to generate phylogenetic tree
2. Pan-genome hierarchically-clustered heatmap: image of clustermap (clustermap.png file) and DataFrame corresponding to the clustermap (PGfam_clustermap_dataframe.txt)
3. Core and unique genes in pan-genome: for each genome, annotation metadata with new column showing which genes are core or unique genes
4. Unique genes in genome subgroups: for each genome, annotation metadata with new column showing which genes are found only in the subgroups specified by the metadata
5. Gene copies versus median gene copy number in genome group: for each genome, annotation metadata with new columns showing how many gene copies are present, the median copy number in all genomes, and if the number genes is equal to, greater than, or equal to the median
6. DIAMOND BLAST: for each genome, annotation metadata with new columns showing which genes are have hits in the virulence factor and antibiotic resistance databases
7. Merged annotation metadata: for each genome, all of the above analyses are merged into a single annotation metadata file

We can use the output from the pan-genome phylogenetic tree (i.e., pan-genome-tree_out.txt) to answer our first question:

• Are E. coli strains from human and rodents phylogenetically distinct?

The pan-genome-tree_out.txt file is a matrix of binary presence and absence of core and accessory genes that is in PHYLIP format. A phylogenetic tree can be created using analyzed by RAxML, IQ-TREE, or similar program. Here, we will download IQ-TREE and then use this program create the phylogenetic tree with the following command:

directory\to\bin\iqtree -s directory\to\pan-genome-tree_out.txt -bb 1000

The output of IQ-TREE, pan-genome-tree_out.txt.tree file, is loaded into FigTree or similar tool to visualize the phylogenetic tree. In the image below, FigTree was used. It appears from the pan-genome phylogenetic tree that the genomes of E. coli strains isolated from rodent hosts are more similar to each other than to the strains from human hosts.

The hierarchically-clustered heatmap (i.e., clustermap) also shows a relationship among the genomes based on the copy number of PGfams. We can use this output to also answer our first question. Based on this clustermap, the E. coli strains from the rodents cluster more closely with each other than the strains from humans.

We also see on the clustermap some obvious differences in the pan-genome among the different E. coli strains. The DataFrame corresponding to the clustermap (PGfam_clustermap_dataframe.txt) can be used to explore these differences if desired, but we will continue using the other outputs from genome_analysis.py to answer the remaining questions.

After each individual comparative analysis step is performed, the results are added to the annotation metadata table for each genome. To facilitate analyzing this data, the results from previous analyses are also merged into a single file, per genome. This makes it easy to examine the datasets and answer the questions we posed above. Below is an excerpt of the merged annotation metadata table for E. coli strain 20170221001. (Note: In the excerpt below, some columns were excluded to optimize viewing.)

genome_name	contig_id	feature_id	type	start	stop	strand	function	pgfam	core_unique_gene	gene_count	median	gene_count-median	vs_median	subgroup_gene	qlen_VF	VF_ID	slen_VF	evalue_VF	bitscore_VF	pident_VF	qcovhsp_VF
Escherichia_coli_strain_20170221001	assembly_contig_1	fig-2.7989.peg.1	CDS	291	1	-	Glutamate decarboxylase (EC 4.1.1.15)	PGF_00008094	core	3	3	0	equal to median
Escherichia_coli_strain_20170221001	assembly_contig_1	fig-2.7989.peg.2	CDS	445	302	-	hypothetical protein	PGF_02969562		1	1	0	equal to median	all_rodents
Escherichia_coli_strain_20170221001	assembly_contig_1	fig-2.7989.peg.3	CDS	472	615	+	hypothetical protein	PGF_10512988	unique_Escherichia_coli_strain_20170221001	1	0	1	greater than median
Escherichia_coli_strain_20170221001	assembly_contig_1	fig-2.7989.peg.4	CDS	1663	653	-	Probable zinc protease pqqL (EC 3.4.99.-)	PGF_09337443	core	1	1	0	equal to median
Escherichia_coli_strain_20170221001	assembly_contig_1	fig-2.7989.peg.5	CDS	3078	1921	-	GALNS arylsulfatase regulator (Fe-S oxidoreductase)	PGF_00006721		1	1	0	equal to median
Escherichia_coli_strain_20170221001	assembly_contig_1	fig-2.7989.peg.6	CDS	4812	3130	-	N-acetylgalactosamine 6-sulfate sulfatase (GALNS)	PGF_00023745		1	1	0	equal to median		560	VFG001444(gb-AAG10151) (aslA) putative arylsulfatase [AslA (VF0238)] [Escherichia coli O18:K1:H7 str. RS218]	475	6.70E-32	136.3	26.1	71.6
Escherichia_coli_strain_20170221001	assembly_contig_1	fig-2.7989.peg.7	CDS	5975	5214	-	Transcriptional regulator YdeO, AraC family	PGF_00058697		1	1	0	equal to median
Escherichia_coli_strain_20170221001	assembly_contig_1	fig-2.7989.peg.8	CDS	6284	6051	-	Two-component-system connector protein SafA	PGF_00063151		1	1	0	equal to median
Escherichia_coli_strain_20170221001	assembly_contig_1	fig-2.7989.peg.9	CDS	8775	6496	-	Putative formate dehydrogenase oxidoreductase protein	PGF_06030909		1	1	0	equal to median
Escherichia_coli_strain_20170221001	assembly_contig_1	fig-2.7989.peg.10	CDS	10028	9114	-	mannose-specific adhesin FimH	PGF_00401044		1	1	0	equal to median		304	VFG042718(gi:15801636) (Z2206) putative adhesin; FimH protein [F9 fimbriae (AI090)] [Escherichia coli O157:H7 str. EDL933]	304	5.80E-171	597.4	98	100
Escherichia_coli_strain_20170221001	assembly_contig_1	fig-2.7989.peg.11	CDS	10591	10088	-	Type 1 fimbrae adaptor subunit FimG	PGF_00063158		1	1	0	equal to median		167	VFG033335(gi:410482685) (fimG) fimbrial-like adhesin protein [Type I fimbriae (CVF426)] [Escherichia coli O104:H4 str. 2009EL-2050]	167	6.80E-89	323.9	98.2	100
Escherichia_coli_strain_20170221001	assembly_contig_1	fig-2.7989.peg.12	CDS	11134	10604	-	Uncharacterized fimbrial-like protein YdeS	PGF_00064710		1	1	0	equal to median		176	VFG033311(gi:387607090) (fimF) fimbrial protein [Type I fimbriae (CVF426)] [Escherichia coli O44:H18 042]	176	1.80E-95	345.9	98.9	100
Escherichia_coli_strain_20170221001	assembly_contig_1	fig-2.7989.peg.13	CDS	13799	11148	-	Outer membrane usher protein FimD	PGF_00028437	core	3	3	0	equal to median		883	VFG033284(gi:387607091) (fimD) fimbrial outer membrane usher protein [Type I fimbriae (CVF426)] [Escherichia coli O44:H18 042]	883	0	1721.1	96.8	100
Escherichia_coli_strain_20170221001	assembly_contig_1	fig-2.7989.peg.14	CDS	14551	13841	-	chaperone FimC	PGF_02911703		1	1	0	equal to median	all_pks	236	VFG042714(gi:15801632) (Z2201) putative fimbrial chaperone protein [F9 fimbriae (AI090)] [Escherichia coli O157:H7 str. EDL933]	239	3.90E-130	461.5	97.9	100
Escherichia_coli_strain_20170221001	assembly_contig_1	fig-2.7989.peg.15	CDS	15476	14913	-	Type-1 fimbrial protein, A chain	PGF_02911992	core	3	3	0	equal to median		187	VFG042713(gi:15801631) (Z2200) putative major fimbrial subunit [F9 fimbriae (AI090)] [Escherichia coli O157:H7 str. EDL933]	187	1.90E-95	345.9	99.5	100
…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…	…

Using the merged annotation metadata, we can answer our remaining questions:

What genes are found in all pks+ E. coli genomes?

To answer this question, we can examine the merged annotation metadata for one of the pks+ E. coli genomes, such as IHE3034. In the “subgroups” column, we can see there are 61 PGfams unique to these pks+ E. coli genomes. Using the virulence factor DIAMOND BLAST result columns, we can these unique genes include pks gene island, which is expected, as well as 14 other virulence factor gene homologs (using a ≥90% sequence identify threshold) including S-fimbriae, Salmochelin siderophore biosynthesis, and the invasion protein IbeA gene.

What genes are found only in genomes of E. coli strains isolated from all rodents?

For this question, let’s examine the merged annotation metadata for one of the E. coli genomes that was isolated from rodent, such as NC101. The “subgroup” shows there are 39 PGfams unique to these seven genomes. Interestingly, 19 of these genes are found in a continuous cluster in the chromosome and are annotated as belonging to propanediol uptake and metabolism. Propanediol can serve as an alternative carbon and energy sources in bacteria encoding these uptake and metabolic pathways, suggesting the E. coli strains isolated from rodents may have different metabolic capabilities compared to reference strains isolated from humans and this could possibly be related to host (i.e., rodent versus human gastrointestinal tract)

What genes are found only in novel genomes of E. coli strains isolated from the rodent?

Now we will examine the merged annotation metadata for one of the novel E. coli strains isolated from the rodent. When we look at E. coli strains 1512290008 as representative, we see there are 15 PGfams unique to this subgroup. These unique genes mainly encode hypothetical protein annotations, and none of the genes had homology to virulence factor or antibiotic resistance genes. These genes are candidates for further investigation. For example, their sequences could be further analyzed with other bioinformatic tools such as Kyoto Encyclopedia of Genes and Genomes (KEGG), InterPro, or NCBI CD-Search to predict protein domains and possibly infer function.

What genes are unique to novel genome of E. coli strains isolated from a rat compared to mice?

For this last question, we will examine the merged annotation metadata for E. coli strains 20170221001, which was isolated from a pet rat. Based on the clustermap, there appears to be a large block of PGfams that are unique to this isolates. Looking at annotation metadata, we see there are 234 PGfams unique to this genome. These unique genomes mainly encode hypothetical protein annotations. Interestingly though, 36 of these unique genes appear to be related to phage genes with 28/36 of these being found continuously on the same contig. It seems the genome from this isolate has a phage present. The sequences of these genes can be further analyzed to identify this phage and its function significance.

Conclusion

This concludes the user documentation for the Microbial Comparative Genomics Workflow. Updates and expanded features will be added to future versions. Please leave comments/feedback here or email them to manniona@mit.edu.

Thank you!

-Tony