PneumoCaT (Pneumococcal Capsular Typing) uses a two-step step approach to assign capsular type to S.pneumoniae genomic data (Illumina). In the first step, reads from each readset are mapped to capsular locus sequences for all known capsular types (92 for S. pneumoniae plus 2 additional subtypes/molecular types). This step is considered successful if the readset matches > 90% to 1 or more capsular locus sequences. If it matches to a single capsular locus then PneumoCaT terminates here and reports this as the assigned capsular type. If more than 1 loci are matched then the tool moves to the second step; a variant based approach that utilises the capsular type variant (CTV) database to distinguish serotypes within a serogroup/genogroup. For more information you can refer to the publication. Please note PneumoCaT applies quality metric requiring a mean depth of 20 reads across the mapped sequence and minimum depth of 5 reads for mapping and will report "Failed" if these conditions are not met.
To keep up to date please watch this repository or follow @PneumoCaT on twitter.
It is very important that the results files produced by PneumoCaT are correctly interpreted. We have become aware of some groups getting poorer than expected concordence in their own validations due to mis-interpretation of the results files. Please read the new result interpretation document (contained in the "Documentation" folder) and also the output files section below carefully to ensure the correct xml file is used as the final result. There are two examples contained in the Examples folder which show the two different result outputs. Please use these to validate that your installation of PneumoCaT is working and also to get familiar with the two different result outputs produced by a serotype which is called only in part 1 (mapping) and a serotype which is called by part 2 (variant based approach).
NOTE 1: For people who are using the published dataset (ENA Project PRJEB14267) to validate the pipeline please download the fastq files under the Submitted Files (FTP) column. Apparently some post-submission processing occuring during ENA submission alters the files causing problems in subsequent analysis.
NOTE 2: For step 2 of the pipeline (but not step 1) there is a requirement for the fastQ files to have a certain name format. The last character of the filename should be the read number (eg ...1.fastq.gz or ....2.fastq.gz.). We will update the method used in the second step to remove this requirement in a future update. If your run is failing after the step 1 mapping while running bowtie in step 2 please check your input filenames.
NOTE 3: V1.2.1 gives an extra determination for serotype 19A and 19A-like 19F pneumococci. Isolates have been found with a capsular operon sequence resembling 19A but that serotype as 19F. In these cases the wzy gene is 19F- like. This determination has been added to the CTVdb in version 1.2.1. Please note it is up to users to validate and decide how to interpret these . Isolates that are resulted as "19AF" by PneumoCaT V1.2.1 should phenotype as 19F but genetically resemble 19A.
Due to the update to new version of Samtools in this version compared to previous version please re-validate outputs as necessary for your quality systems.
- Dependencies
- Running PneumoCaT
- PneumoCaT Output
- CTV database
- Examples
- Troubleshooting
- Contact Information
- Citing PneumoCaT
- Licence Agreement
PneumoCaT is written with Python 2.7.5 and requires the following packages installed before running:
- Bowtie2 (https://sourceforge.net/projects/bowtie-bio/files/bowtie2/2.2.9/)
- Samtools version 1.7 (https://sourceforge.net/projects/samtools/files/samtools/1.7/) (see NOTE below). To install after downloading:
cd <path-to>/samtools-1.7
make
- PyYaml (http://pyyaml.org/)
- numpy (http://www.scipy.org/scipylib/download.html)
- lxml (http://lxml.de/installation.html)
- pysam (https://github.com/pysam-developers/pysam)
- biopython (https://github.com/biopython/biopython)
NOTE: Previous dependency to Samtools version 0.1.19 was due to incompatibility of pysam with version 1.2. This was resolved with version 1.3 so PneumoCaT has been changed to work with this version of Samtools and hopefully the ones after that (versions 1.3+). Version 0.1.19 is no longer compatible due to changes on the structure of the samtools command specifically, the latest versions require an output filename using tag -o instead of the prefix required by the earliest versions.
version 0.1.19
samtools sort [options] <in.bam> <out.prefix>
version 1.3+
samtools sort [options...] -o FILENAME <in.bam>
For step by step instructions for installing all dependencies using conda see the SetUpInstructions.pdf document available in Documentation.
usage: python PneumoCaT.py [-h] [--input_directory INPUT_DIRECTORY]
[--fastq_1 FASTQ_1] [--fastq_2 FASTQ_2]
[--variant_database VARIANT_DATABASE]
[--output_dir OUTPUT_DIR] [--threads THREADS]
[--bowtie BOWTIE] [--samtools SAMTOOLS] [--cleanup]
optional arguments:
-h, --help show this help message and exit
--input_directory INPUT_DIRECTORY, -i INPUT_DIRECTORY
please provide the path to the directory that contains the
fastq files [REQUIRED - OPTION 1]
--fastq_1 FASTQ_1, -1 FASTQ_1
Fastq file pair 1 [REQUIRED - OPTION 2]
--fastq_2 FASTQ_2, -2 FASTQ_2
Fastq file pair 2 [REQUIRED - OPTION 2]
--variant_database VARIANT_DATABASE, -d VARIANT_DATABASE
variant database [OPTIONAL]; defaults to
streptococcus-pneumoniae-ctvdb in the github directory
--output_dir OUTPUT_DIR, -o OUTPUT_DIR
please provide an output directory [OPTIONAL]; if none
provided a pneumo_capsular_typing folder will be
created in the directory containing the fastq files
--threads THREADS, -t THREADS
number of threads to use [OPTIONAL];
default=4
--bowtie BOWTIE, -b BOWTIE
please provide the path for bowtie2 [OPTIONAL];
defaults to bowtie2
--samtools SAMTOOLS, -s SAMTOOLS
please provide the path for samtools [OPTIONAL];
defaults to samtools
--cleanup, -c if used, all bam files generated will be removed upon
completion
Input files are required to have this pattern present *1.fastq* or *2.fastq*. The tool doesn't currently recognise *fq*. SAMPLEID is extracted from the 1.fastq file by splitting the fastq file name on '.' and selecting the first part, i.e. SAMPLEID from SAMPLEID.1.fastq.gz or SAMPLEID_1 from SAMPLEID_1.fastq.gz.
- SAMPLEID.results.xml - The XML file at step 1 is basic and it only contains the top two capsular types and their respective read coverage (% of the capsular locus length covered). PLEASE NOTE this file ONLY contains the final result if no "SNP_based_serotyping" folder appears in the result output location.
If only one capsular type is matched with more than 90% coverage then the report from step 1 contained in this xml file is considered the final result (result type="Serotype") and no further folders will appear within the PneumoCaT output folder. If more than one capsular type are matched with more than 90% coverage then the software moves to step two and a SNP_based_serotyping folder is created containing a second XML file with the final result - see STEP 2- VARIANT-BASED APPROACH.
Note that the output XML file from step 1 only reports two capsular types, when actually more could be matched and all will pass to step 2 for further distinction. Further information on mapped serotypes in stage 1 can be found in "Coverage_summary.txt". If the top hit coverage is < 90% then no serotypes are reported and 'Failed' appears instead.
- SAMPLEID.sorted.bam - BAM file generated during step 1 using the 94 capsular locus sequences as reference.
- SAMPLEID.sorted.bam.bai - index file for the sorted BAM file
- ComponentComplete.txt - indicates PneumoCaT analysis was completed succssfully - useful if using PneumoCaT in a pipeline.
- coverage_summary.txt - contains the coverage values for all serotypes. This is useful if the step has failed, epsecially if the top coverage falls close to the 90% threshold.
- SAMPLEID.results.xml - The XML file at step 2 details the final assigned capsular type (result type ="Serotype Distinction"), total hits, the capsular types studied in this analysis as well as a full breakdown of the variant positions used for this assignment. Total hits corresponds to the number of matched variants vs the number of all variants tested. A capsular type is assigned only if all variants matched. If a complete match is not possible then 'Serotype Undetermined' is reported. Coverage statistic metrics are calculated for each gene locus used for this assignment. These include five values as detailed below:
- Minimum depth: the minimum number of reads that maps to the gene sequence at any single position.
- Maximum depth: the maximum number of reads that maps to the gene sequence at any single position.
- Average depth: the average number of reads that maps across the length of the gene sequence.
- Percent Coverage: records coverage across the length of each gene sequence.
- Coverage Distribution: records the area of the gene sequence covered by reads.
- SAMPLEID.sorted.bam - BAM file generated during step 2 using specific genes, assigned for each serogroup/genogroup in the CTV database, as reference
- SAMPLEID.sorted.bam.bai - index file for the sorted BAM file
- variant_summary.yml - contains the variant matches for all serotypes tested. This is useful if a complete match was not possible and 'Serotype undetermined' was reported. It can give information on coverage for specific positions, mixed positions and other events that could lead to failure to assing a serotype.
Capsular Type Variant (CTV) database is supplied as a structured directory (streptococcus-pneumoniae-ctvdb). Within this directory you can find:
- reference.fasta - This file contains the capsular locus sequences for all 92 serotypes plus 6E and 23B1 first described in the publication.
- 19 subfolders corresponding to each genogroup recognised in publication. Each subfolder contains:
- mutationdb.pickled - contains the variants that can distinguish the serotypes within this serogroup. This file is read by PneumoCaT during step 2.
- mutationdb.yml - contains the same information as the mutationdb.pickled file but can be read using text editors.
- reference.fasta - contains the sequences for the genes defined in the mutationdb file.
For users that want to specify their own database of reference capsular locus sequences please follow the following instructions:
-
Create a copy of streptococcus-pneumoniae-ctvdb in the PneumoCaT directory
cp -r streptococcus-pneumoniae-ctvdb streptococcus-pneumoniae-custom -
Within the streptococcus-pneumoniae-custom directory open the reference.fasta file and add/remove reference capsular locus sequences.
-
Make sure to save the file with the same name (reference.fasta).
-
Alternatively, if the user has an alternative fasta file already available, remove the reference.fasta file from streptococcus-pneumoniae-custom and replace it with the alternative fasta file. Again make sure to change the name to reference.fasta
-
When running PneumoCaT define the new database using the option -d followed by the path to streptococcus-pneumoniae-custom (see below).
Default for all optional arguments:
python PneumoCaT.py -i <path-to-input-directory>
* This assumes that the paths to bowtie2 and samtools are included in the PATH environment variable. If not then options -b and -s should be used to define the full path to bowtie2 and samtools, respectively (see below).
Example:
If the PneumoCaT directory was cloned to your environment follow these commands to run example PHESPV0253 using the provided reference database:
cd PneumoCaT
python PneumoCaT.py -i Examples/PHESPV0253
* This defaults options -b and -s to bowtie2 and samtools and will only work if the respective paths are added into the PATH environment database.
With Optional arguments:
python PneumoCaT.py -i <path-to-input-directory> -o <path-to-output-directory> -d <path-to-the-database> -b <path-to-bowtie2> -s <path-to-samtools>
Example:
If the PneumoCaT directory was cloned to your environment follow these commands to run example PHESPV0253 using a custom reference database and with defined paths to bowtie2 and samtools:
cd PneumoCaT
python PneumoCaT.py -i Examples/PHESPV0253 -d streptococcus-pneumoniae-custom -b /usr/local/bowtie2-2.2.9/bowtie2 -s /usr/local/samtools-0.1.19/samtools
Default for all optional arguments:
python PneumoCaT -1 <path-to-FASTQ1> -2 <path-to-FASTQ2>
Example:
If the PneumoCaT directory was cloned to your environment follow these commands to run example PHESPV0253 :
cd PneumoCaT
python PneumoCaT.py -1 Examples/PHESPV0253/PHESPV0253.R1.fastq.gz -2 Examples/PHESPV0253/PHESPV0253.R2.fastq.gz
With Optional arguments:
python PneumoCaT -1 <path-to-FASTQ1> -2 <path-to-FASTQ2> -o <path-to-output-directory> -d <path-to-the-database> -b <path-to-bowties> -s <path-to-samtools>
As mentioned above, serotype can be called either in STEP 1 when a highly distinctive capsular locus sequence is present or STEP 2 when two or more serotypes have a similar capsular locus sequence and the variants from CTV database are used for distinction. In Examples folder, an example for each of the scenarios is given; PHESPV0253 isolate is serotype 7B by both WGS and slide agglutination and it's an example of PneumoCaT STEP 2 analysis, whereas PHESPV1910 is serotype 5 and is an example of PneumoCaT STEP 1 analysis. Each folder contains the fastq files and the expected final result.xml file.
Five different formats of failure values are returned when no serotype can be predicted:
- "Reference data not available: Serogroup 24": Serogroups 24 and 32 have no available reference data.
- "Mixed: ['01', '22A', '22F']": The combination of serotypes matched with more than 90% coverage during step 1 does not correspond to any known serogroup or genogroup.In this case mixed culture is suspected.
- "Mixed: ['06A', '06B']": If mixed is returned at step 2 then a "Mixed" tag has been returned for one or more of the variants. For example:
- "Mixed: {'allele-2': 0.99, 'allele-1': 0.99}": Mixed samples based on presence of two alleles for the gene with breakdown of coverage for the alleles found. There could be two scenarios in this case:
- {'allele-2': 0.99, 'allele-1': 0.99}: when both alleles has coverage more than 90% then a mixed culture is suspected.
- {'allele-1': 0.75, 'allele-2': 0.25}: In this case, further bioinformatics investigation is required to determine chimeric state.
- "Mixed: (721, 722),G/-,G:0.67,-:0.33”: Mixed genotype at specific region with breakdown of frequency for each state. Usually corresponds to frameshift mutation responsible for loss of function (pseudogene).
- "Mixed: (215: {'A': 0, 'C': 64, 'T': 67, 'G': 0})”: Mixed genotype at specific position with breakdown of nucleotides found.
- "Mixed: {'allele-2': 0.99, 'allele-1': 0.99}": Mixed samples based on presence of two alleles for the gene with breakdown of coverage for the alleles found. There could be two scenarios in this case:
- "Serotype undetermined": If no variant pattern is matched 100% then no serotype can be predicted and "Serotype undetermined" is reported. This could be due to either a new variant pattern that should be investigated further (e.g. new molecular subtype or new serotype) or coverage issues:
- “Low coverage”: Less than 90% coverage of the gene sequence.
- "SNP position 1002: Depth < 5”: Nucleotide position covered by less than 5 reads.
- “SNP position 1002 not covered”: No reads map at this nucleotide position.
- "Failed": This output is associated with step 1 and coverage <90% for the capsular locus sequence with the highest coverage OR when the minimum read depth is <5 or average read depth is <20. This tag either denote an absence of a fully functional capsular locus ( can be interpreted in two different ways based on the coverage values as below) OR that the quality of sequencing data has not passed the threshold values for analysis.
If the depth metrics are above the thresholds and the result = Failed:
- If coverage < 60%: this corresponds to a non-typable isolate. This was tested in the lab with clinical isolates using a traditional serotyping method (slide agglutination with SSI serum). This will be tested further and an appropriate result should be added in the future.
- If coverage > 60%: this corresponds to a partial capsule and in all cases seen so far the isolate still expresses a capsule.
- 15B/C: A presence of a mixed 15B/C profile ("Mixed: (413, 414),TA/--,TA:0.67,--:0.33”) in wciZ gene has been reported in various publications and 15B/C is now widely accepted as a reported serotype and is a valid output for PneumoCaT. However "Mixed: ['15B', '15C']" might still be reported and this should not be confused with 15B/C. In this case, "Mixed" has been detected in variants other than wciZ pos 413 and this should be investigated further.
- "Mixed: ['07A', '07F']": corresponds to a serotype 7A isolate. Similar to 15B/C, 7A and 7F are distinguished by a frameshift mutation and as yet no pure 7A (wcwD insT 587) has been found using PneumoCaT (n=3). In the cases where mixed profile was given they serotyped as 7A. Unlike 15B/C this has not been validated in a large number of samples so no inferences can yet be made. We would be grateful for any serotype 7A or Mixed: ['07A', '07F'] to help with further PneumoCaT development.
PneumoCaT applies a minimum read depth of 5 and minimum average read depth of 20 for mapping. Samples that do not meet these criteria are scored as "Failed" even when the coverage is more than 90%. These threshold values could be modified in the serotype_determiner_functions script if lower stringency is required.
Variants can confidently reported if the percentage coverage for the gene is more than 90% and a single fragment is reported in the coverage distribution. As well as this general threshold individual mutations have specific thresholds.
- Alleles: An allele is reported if it has coverage more than 90% and the other alleles have coverage < 10%, otherwise a mixed tag is reported.
- SNPs: When a specific nt position is investigated the hit will only be reported if
- depth in the codon area is more than 5
- consensus base has more than 75% frequency
- mapping quality is more than 30
- no other snps are reported within ±5 nts area
- Pseudogenes: The overall coverage threshold has been reduced to 70% to allow for reporting large indels/truncations as a causal agent for loss of function. A functional orf is reported if coverage is more than 80% and a single orf is present that match the functional orf for the gene tested. This lower threshold (80% instead of 90%) has been empirically adjusted to allow for poor mapping in the beginning and end of the gene locus. If specific frameshift mutations are investigated then the same threshold as described above are used (SNPs).
Please cite the original paper:
Whole genome sequencing of Streptococcus pneumoniae: development, evaluation and verification of targets for serogroup and serotype prediction using an automated pipeline
Georgia Kapatai, Carmen L. Sheppard, Ali Al-Shahib, David J. Litt, Anthony P. Underwood, Timothy G. Harrison, Norman K. Fry
PeerJ: 2016 Sep 14;4:e2477. doi: 10.7717/peerj.2477. eCollection 2016.
This software is covered by GNU General Public License, version 3 (GPL-3.0).