Recipe View

This recipe evaluates the accuracy of the Centrifuge classification for selected markers.

2 results • updated 2.7 years ago by Istvan Albert

This recipe evaluates the accuracy of the Centrifuge classification for selected markers.

The recipe generates simulated sequencing reads from NCBI accession numbers then performs classification with Centrifuge

In the third step evaluates and reports the accuracy (the percent of reads were classified correctly) of the classification.

The input to the recipe is a list of NCBI accession numbers:

AP012081
AP011279
DQ288268
AP014537

The main output is a file that lists the classification accuracy for each accession number:

accession     expect    actual   percent   taxid      title
AP012081      1000      993      99.3      67547      Margariscus margarita mitochondrial DNA, almost complete genome, except for D-loop
AP011279      1000      996      99.6      90988      Pimephales promelas mitochondrial DNA, complete genome
DQ288268      1000      996      99.6      8022       Oncorhynchus mykiss mitochondrion, complete genome
AP014537      1000      1000    100.0      27706      Micropterus salmoides mitochondrial DNA, almost complete genome

Copy recipe

You need write access to the original recipe to edit.

set -uex


# Input file that contains the accession numbers.
ACC=FILENAME


# Use the taxonomy specific files to build the custom database.
TAXDIR=~/refs/taxonomy

# Files used in generating the Centrifuge index.
TABLE=$TAXDIR/table.txt
NODES=$TAXDIR/nodes.dmp
NAMES=$TAXDIR/names.dmp

# Make the taxonomy directory.
mkdir -p $TAXDIR

#
# Download the taxonomy for centrifuge.
# Needs to be done separately and only once.
#

#
# Prepare the taxonomy data.
# This operation only needs to be done once for the entire website.
#
# Download the taxonomy files from NCBI.
# (cd $TAXDIR && wget ftp://ftp.ncbi.nlm.nih.gov/pub/taxonomy/taxdump.tar.gz)
# (cd $TAXDIR && wget ftp://ftp.ncbi.nih.gov/pub/taxonomy/accession2taxid/nucl_gb.accession2taxid.gz)

# Uncompress the taxonomy files.
# (cd $TAXDIR &&  tar -xzvf $TAXDIR/taxdump.tar.gz)
# (cd $TAXDIR && gunzip nucl_gb.accession2taxid.gz)

# Create the conversion table (accession to taxid mapping).
# cat $TAXDIR/nucl_gb.accession2taxid | cut -f 2,3 > $TABLE


# How many reads to simulate.
N=1000

# Lenght of reads.
L=100

# Read error rates.
E=0.02

# Will store data in these locations
mkdir -p input output code

# The FASTA file storing the sequence for accession numbers.
FASTA=input/reference.fa

# The FASTQ file storing the reads simulated from the reference.
R1=input/r1.fq
R2=input/r2.fq

# The accesion taxonomy file
TAXONOMY=input/taxonomy.txt

# The Centrifuge classification report.
REPORT=output/centrifuge-report.txt

# The Centrifuge output.
OUTPUT=output/centrifuge-output.txt

# The Centrifuge accuracy report.
ACCURACY=accuracy-centrifuge.txt

# Copy over the accession file to input folder to aid reproducibility.
cp $ACC input/accessions.txt

# Get sequences from NCBI.
epost -db nuccore -input $ACC | efetch -format fasta > $FASTA

# Identify the taxonomy for each accession number.
epost -db nuccore -input $ACC  | esummary | xtract -pattern DocumentSummary -element Caption,TaxId,Title > $TAXONOMY

# Store the Centrifuge database under the checksum name.
INDEX=index/db

# Make thr directory
mkdir -p index

# Build the centrifuge index.
centrifuge-build -p 2 --conversion-table $TABLE --taxonomy-tree $NODES --name-table $NAMES $FASTA $INDEX >> runlog.txt

# Install the plac command parser.
pip install plac -q

# Clone the read simulator. Must use wgsim2.
git clone --depth=1 https://github.com/ialbert/wgsim2.git > log.txt

# Compile the read simulator.
(cd wgsim2 && make 2> log.txt)

# Simulate reads with errors in fixed N mode.
wgsim2/wgsim2 -f -r 0 -R 0 -N $N -e $E -1 $L -2 $L $FASTA $R1 $R2 > log.txt

# Run Centrifuge
centrifuge -x $INDEX -1 $R1 -2 $R2 -S $OUTPUT  --report-file $REPORT

# Kraken style report
centrifuge-kreport -x $INDEX $OUTPUT > $REPORT

# The location of the code that validates the reads.
URL2=https://raw.githubusercontent.com/biostars/biocode/master/scripts/classify/validate.py

# Get the code.
curl $URL2 > code/validate.py

# Run the validator
python code/validate.py -f $OUTPUT -t $TAXONOMY -c $N > $ACCURACY

You need write access to the original recipe to edit.

Click the buttons on the right to create new fields.

Add text field Add float field Add data field Add checkbox Add dropdown Add upload field Add integer field Add radio button

Name

Recipe display name

Identifier

Unique identifier for the recipe.

Image :

Optional image for the recipe ( 500px Maximum ).

Rank:

Used to order recipes (optional).

This recipe evaluates the accuracy of the Centrifuge classification for selected markers.

The recipe generates simulated sequencing reads from NCBI accession numbers then performs classification with Centrifuge

In the third step evaluates and reports the accuracy (the percent of reads were classified correctly) of the classification.

The input to the recipe is a list of NCBI accession numbers:

```
AP012081
AP011279
DQ288268
AP014537
```

The main output is a file that lists the classification accuracy for each accession number:

```
accession     expect    actual   percent   taxid      title
AP012081      1000      993      99.3      67547      Margariscus margarita mitochondrial DNA, almost complete genome, except for D-loop
AP011279      1000      996      99.6      90988      Pimephales promelas mitochondrial DNA, complete genome
DQ288268      1000      996      99.6      8022       Oncorhynchus mykiss mitochondrion, complete genome
AP014537      1000      1000    100.0      27706      Micropterus salmoides mitochondrial DNA, almost complete genome
```

Insert Image

From the web

From your computer

Cancel

Back

Centrifuge classification validator

Insert Image