Merge pull request #357 from lskatz/lskatz-patch-1

Lskatz patch 1

Program_Licenses.md

@@ -26,6 +26,7 @@ The licenses of the open-source software that is contained in these Docker image
 | colorid | MIT | https://github.com/hcdenbakker/colorid/blob/master/LICENSE |
 | DSK | GNU Affero GPLv3 | https://github.com/GATB/dsk/blob/master/LICENSE |
 | emm-typing-tool | GNU GPLv3 | https://github.com/phe-bioinformatics/emm-typing-tool/blob/master/LICENCE |
+| EToKi | GNU GPLv3 | https://github.com/zheminzhou/EToKi/blob/master/LICENSE |
 | FastANI | Apache v2.0 | https://github.com/ParBLiSS/FastANI/blob/master/LICENSE |
 | FastTree | GNU GPLv2 | http://www.microbesonline.org/fasttree/ |
 | FastQC | GNU GPLv3 | https://github.com/s-andrews/FastQC/blob/master/LICENSE.txt |

README.md

@@ -57,6 +57,7 @@ To learn more about the docker pull rate limits and the open source software pro
 | [datasets-sars-cov-2](https://github.com/CDCgov/datasets-sars-cov-2) <br/> [![docker pulls](https://badgen.net/docker/pulls/staphb/datasets-sars-cov-2)](https://hub.docker.com/r/staphb/datasets-sars-cov-2) | <ul><li>0.6.2</li><li>0.6.3</li></ul> | https://github.com/CDCgov/datasets-sars-cov-2 |
 | [DSK](https://hub.docker.com/r/staphb/dsk) <br/> [![docker pulls](https://badgen.net/docker/pulls/staphb/dsk)](https://hub.docker.com/r/staphb/dsk) | <ul><li>0.0.100</li></ul> | https://gatb.inria.fr/software/dsk/ |
 | [emm-typing-tool](https://hub.docker.com/r/staphb/emm-typing-tool) <br/> [![docker pulls](https://badgen.net/docker/pulls/staphb/emm-typing-tool)](https://hub.docker.com/r/staphb/emm-typing-tool) | <ul><li>0.0.1 (no version)</li></ul> | https://github.com/phe-bioinformatics/emm-typing-tool |
+| [EToKi](https://hub.docker.com/r/staphb/etoki) <br/> [![docker pulls](https://badgen.net/docker/pulls/staphb/etoki)](https://hub.docker.com/r/staphb/etoki) | <ul><li>1.2.1</li></ul> | https://github.com/zheminzhou/EToKi |
 | [FastANI](https://hub.docker.com/r/staphb/fastani) <br/> [![docker pulls](https://badgen.net/docker/pulls/staphb/fastani)](https://hub.docker.com/r/staphb/fastani) | <ul><li>1.1</li><li>1.32</li><li>1.33</li></ul> | https://github.com/ParBLiSS/FastANI |
 | [FastTree](https://hub.docker.com/r/staphb/fasttree) <br/> [![docker pulls](https://badgen.net/docker/pulls/staphb/fasttree)](https://hub.docker.com/r/staphb/fasttree) | <ul><li>2.1.11</li></ul> | http://www.microbesonline.org/fasttree/ |
 | [FastQC](https://hub.docker.com/r/staphb/fastqc) <br/> [![docker pulls](https://badgen.net/docker/pulls/staphb/fastqc)](https://hub.docker.com/r/staphb/fastqc) | <ul><li>0.11.8</li><li>0.11.9</li></ul> | https://www.bioinformatics.babraham.ac.uk/projects/fastqc/ <br/> https://github.com/s-andrews/FastQC |

etoki/1.2.1/Dockerfile

@@ -0,0 +1,219 @@
+# FYI this is a multistage build, which can get very complicated
+
+ARG ETOKI_VER="1.2.1"
+ARG SKESA_VER=2.4.0
+ARG SPADES_VER=3.15.4
+ARG SHOVILL_VER=1.1.0
+ARG KRAKEN2_VER=2.1.2-no-db
+ARG BOWTIE_VER=2.4.4
+ARG LYVESET_VER=1.1.4f
+#ARG VSEARCH_VER=2.21.1
+ARG BBTOOLS_VER=38.96
+#ARG MLST_VER=2.19.0
+ARG ORTHOFINDER_VER=2.17
+ARG CFSAN_VER=2.0.2
+ARG FLYE_VER=2.9
+ARG SAMTOOLS_VER=1.15
+ARG DIAMOND_VER="v2.0.15"
+ARG RAXML_VER="8.2.12"
+
+FROM staphb/shovill:${SHOVILL_VER}          AS shovill
+FROM staphb/kraken2:${KRAKEN2_VER}          AS kraken2
+FROM staphb/bowtie2:${BOWTIE_VER}           AS bowtie2
+FROM staphb/lyveset:${LYVESET_VER}          AS lyveset
+FROM staphb/bbtools:${BBTOOLS_VER}          AS bbtools
+#FROM staphb/mlst:${MLST_VER}                AS mlst
+#FROM staphb/orthofinder:${ORTHOFINDER_VER}  AS orthofinder
+FROM staphb/cfsan-snp-pipeline:${CFSAN_VER} AS cfsan
+FROM staphb/flye:${FLYE_VER}                AS flye
+FROM staphb/samtools:${SAMTOOLS_VER}        AS samtools
+FROM staphb/raxml:${RAXML_VER}              AS raxml
+
+FROM ubuntu:jammy as app
+
+LABEL base.image="ubuntu:jammy"
+LABEL dockerfile.version="1"
+LABEL software="EToKi"
+LABEL software.version=$ETOKI_VER
+LABEL description="All methods related to Enterobase data analysis pipelines"
+LABEL website="https://github.com/zheminzhou/EToKi"
+LABEL license="https://github.com/zheminzhou/EToKi/blob/master/LICENSE"
+LABEL maintainer="Lee Katz"
+LABEL maintainer.email="[email protected]"
+
+ARG DEBIAN_FRONTEND=noninteractive
+
+RUN apt-get update && apt-get install -y --no-install-recommends \
+ libncurses5-dev \
+ libbz2-dev \
+ liblzma-dev \
+ perl-base \
+ libcurl4-gnutls-dev \
+ gcc \
+ g++ \
+ python-setuptools \
+ zlib1g-dev \
+ libbz2-dev \
+ python-is-python3 \
+ python3-pip \
+ python3-dev \
+ libgconf-2-4 \
+ curl \
+ unzip \
+ build-essential \
+ git \
+ pigz \
+ libcurl4 \
+ ant \
+ libssl-dev \
+ python3-venv \
+ autoconf \
+ automake \
+ make \
+ wget \
+ mmseqs2 \
+ ncbi-blast+ && \
+ apt-get autoclean && \
+ rm -rf /var/lib/apt/lists/*
+
+RUN pip3 install ete3 numba pandas scikit-learn psutil click scipy numpy==1.21.6
+
+COPY --from=shovill     /skesa/skesa            /usr/local/bin/
+COPY --from=shovill     /megahit                /megahit
+COPY --from=shovill     /pilon                  /pilon
+COPY --from=shovill     /SPAdes-*-Linux         /spades
+COPY --from=kraken2     /kraken2-2*             /kraken2
+COPY --from=kraken2     /kraken2-db             /kraken2-db
+COPY --from=bowtie2     /opt/bowtie2-*          /opt/bowtie2
+COPY --from=bbtools     /opt/bbmap              /opt/bbmap
+#COPY --from=mlst        /ncbi-blast-2.9.0+      /ncbi-blast-2.9.0+
+COPY --from=lyveset     /lyve-SET               /lyve-SET
+#COPY --from=orthofinder /mmseqs                 /mmseqs
+COPY --from=cfsan       /gatk                   /gatk
+COPY --from=flye        /Flye-*                 /flye
+COPY --from=samtools    /samtools-*             /samtools
+COPY --from=raxml       /standard-RAxML-8.2.12  /standard-RAxML
+COPY --from=raxml       /raxml_ng               /raxml_ng
+
+ARG DIAMOND_VER
+ARG ETOKI_VER
+#ARG VSEARCH_VER
+
+# Diamond
+RUN wget https://github.com/bbuchfink/diamond/releases/download/${DIAMOND_VER}/diamond-linux64.tar.gz && \
+ tar zxvf diamond-linux64.tar.gz && \
+ mv -v diamond /usr/local/bin/diamond && \
+ chmod +x /usr/local/bin/diamond && \
+ rm diamond-linux64.tar.gz
+
+# mmseqs
+#RUN wget https://mmseqs.com/latest/mmseqs-linux-avx2.tar.gz && \
+#RUN wget --no-check-certificate --secure-protocol=TLSv1_2 --debug -v --auth-no-challenge http://mmseqs.com/latest/mmseqs-linux-avx2.tar.gz && \
+#  tar -xvf mmseqs-linux-avx2.tar.gz && \
+#  rm -rf mmseqs-linux-avx2.tar.gz && \
+#  /bin/bash -c "source /mmseqs/util/bash-completion.sh"
+
+
+ENV PATH="/usr/local/bin:/EToKi-${ETOKI_VER}:/flye/bin:/megahit:/pilon:/spades/bin:/kraken2:/opt/bowtie2:/opt/bbmap:/ncbi-blast-2.9.0+/bin:/megahit/megahit_v1.1.4_LINUX_CPUONLY_x86_64-bin:/raxml_ng:/standard-RAxML:/lyve-SET:/lyve-SET/scripts:/mmseqs/bin:/mmseqs/util:$PATH" \
+   LC_ALL=C
+
+# Test many executables
+#RUN mmseqs --version
+#RUN flye --version
+#RUN samtools --help
+#RUN diamond --help && diamond --version
+
+# Vsearch sort of copied from https://github.com/torognes/vsearch/blob/v2.21.1/Dockerfile
+#WORKDIR /opt
+#RUN git clone https://github.com/torognes/vsearch.git && cd vsearch && git checkout v${VSEARCH_VER}
+#WORKDIR /opt/vsearch
+#RUN ./autogen.sh && \
+    #./configure CFLAGS="-O3" CXXFLAGS="-O3" && \
+    #make clean && \
+    #make && \
+    #make install && \
+    #make clean && \
+    #cd .. && \
+    #rm -rf /opt/vsearch
+
+WORKDIR /
+
+## EToKi itself ##
+#RUN wget https://github.com/zheminzhou/EToKi/archive/refs/tags/${ETOKI_VER}.tar.gz && \
+#  ls / && \
+#  tar zxvf ${ETOKI_VER}.tar.gz && \
+#  rm ${ETOKI_VER}.tar.gz
+# TODO checkout a specific tag or hashtag
+RUN wget https://github.com/lskatz/EToKi/archive/refs/tags/${ETOKI_VER}.tar.gz && \
+ tar zxvf ${ETOKI_VER}.tar.gz
+
+#RUN git clone https://github.com/lskatz/EToKi.git /EToKi-${ETOKI_VER} && \
+# cd /EToKi-${ETOKI_VER} && \
+# git checkout 1.2.1 && \
+# cd -
+
+WORKDIR /EToKi-${ETOKI_VER}
+
+RUN EToKi.py configure --help
+
+RUN EToKi.py configure --path bbduk=/opt/bbmap/bbduk.sh 
+RUN EToKi.py configure --path bbmerge=$(which bbmerge.sh) 
+RUN EToKi.py configure --path bbduk=/opt/bbmap/bbduk.sh 
+RUN EToKi.py configure --path bbmerge=$(which bbmerge.sh) 
+RUN EToKi.py configure --path repair=$(which repair.sh) 
+RUN EToKi.py configure --path pilon=$(find /pilon -name 'pilon-*.jar' |  head -n 1) 
+RUN EToKi.py configure --path flye=$(which flye) 
+RUN EToKi.py configure --path kraken2=$(find /kraken2 -type f -name kraken2 | head -n 1) 
+RUN EToKi.py configure --path bowtie2=$(which bowtie2) 
+RUN EToKi.py configure --path bowtie2build=$(which bowtie2-build) 
+RUN EToKi.py configure --path raxml=$(which raxmlHPC) 
+RUN EToKi.py configure --path raxml_ng=$(which raxml-ng) 
+RUN EToKi.py configure --path samtools=$(which samtools) 
+RUN EToKi.py configure --path blastn=$(which blastn) 
+RUN EToKi.py configure --path blastp=$(which blastp) 
+RUN EToKi.py configure --path makeblastdb=$(which makeblastdb) 
+RUN EToKi.py configure --path diamond=$(which diamond) 
+RUN EToKi.py configure --path gatk=/gatk/GenomeAnalysisTK-3.8-1-0-gf15c1c3ef/GenomeAnalysisTK.jar 
+RUN EToKi.py configure --path mmseqs=$(which mmseqs) 
+RUN EToKi.py configure --path megahit=$(which megahit) 
+RUN EToKi.py configure --path spades=$(which spades.py) 
+# In the LK version, we are emulating usearch with blastp
+RUN EToKi.py configure --usearch $(which blastp)
+
+WORKDIR /data
+
+FROM app as test
+
+ARG ETOKI_VER
+
+WORKDIR /EToKi-${ETOKI_VER}
+# Get the help menu up
+RUN EToKi.py --help
+# Show the configuration works
+RUN EToKi.py configure  
+
+ENV MLSTdb="/EToKi-${ETOKI_VER}/examples/Escherichia.Achtman.alleles.fasta"
+ENV MLSTalleles="/EToKi-${ETOKI_VER}/examples/Escherichia.Achtman.alleles.fasta"
+ENV MLSTtab="/EToKi-${ETOKI_VER}/examples/Escherichia.Achtman.convert.tab"
+ENV ECOLI_assembly="/EToKi-${ETOKI_VER}/examples/GCF_000005845.2_ASM584v2_genomic.fna"
+ENV ECOLI_assembly2="/EToKi-${ETOKI_VER}/examples/GCF_000214765.2_ASM21476v3_genomic.fna"
+ENV ECOLI_assembly3="/EToKi-${ETOKI_VER}/examples/GCF_001566635.1_ASM156663v1_genomic.fna"
+RUN gunzip -vf ${ECOLI_assembly} ${ECOLI_assembly2} ${ECOLI_assembly3}
+
+# Try out MLST
+RUN EToKi.py MLSTdb -i ${MLSTalleles} -r ${MLSTdb} -d ${MLSTtab}
+# Show some results of the database creation
+RUN ls -lh $MLSTdb $MLSTalleles $MLSTtab && head ${MLSTalleles} ${MLSTdb} ${MLSTtab}
+# Run typing
+RUN EToKi.py MLSType -i ${ECOLI_assembly3} -r ${MLSTdb} -k G749 -o stdout -d ${MLSTtab}
+
+# Return the fasta files back to gzip status
+RUN gzip -vf ${ECOLI_assembly} ${ECOLI_assembly2} ${ECOLI_assembly3}
+
+# Just take the examples of unit tests that are most relevant to MLST
+# from example.bash in the repo
+#RUN python EToKi.py MLSTdb -i examples/Escherichia.Achtman.alleles.fasta -r examples/Escherichia.Achtman.references.fasta -d examples/Escherichia.Achtman.convert.tab
+#RUN gzip -cd examples/GCF_001566635.1_ASM156663v1_genomic.fna.gz > examples/GCF_001566635.1_ASM156663v1_genomic.fna && \
+
+
+