Just teraseq

teraseq/Dockerfile

@@ -1,8 +1,212 @@
-FROM ghcr.io/binpash/teraseq20:latest
+# Download base image ubuntu 16.04
+FROM ubuntu:16.04
+
+# LABEL about the custom image
+LABEL maintainer="[email protected]"
+LABEL version="0.3"
+LABEL description="This is custom Docker Image for \
+analysis of TERA-Seq publication (DOI: https://doi.org/10.1093/nar/gkab713)."
+
+# Disable Prompt During Packages Installation
+ARG DEBIAN_FRONTEND=noninteractive
+
+# Set default shell
+SHELL ["/bin/bash", "-c"]
+
+### System-wide requirements; cpanminus is not required if Perl uses virtual environment method; g++, zlib1g-dev, and bzip2 are required only for Nanopolish
+RUN apt-get update \
+    && apt-get install -y git gcc make wget g++ zlib1g-dev bzip2 \
+    && rm -rf /var/lib/apt/lists/*
+
+### Main GitHub repo
+WORKDIR /root
+COPY . TERA-Seq_manuscript
+
+### Install Miniconda3
+ENV PATH "/root/miniconda3/bin:${PATH}"
+ARG PATH="/root/miniconda3/bin:${PATH}"
+
+RUN wget \
+    https://repo.anaconda.com/miniconda/Miniconda3-py37_23.1.0-1-Linux-x86_64.sh -O Miniconda3.sh \
+    && mkdir /root/.conda \
+    && bash Miniconda3.sh -b \
+    && rm -f Miniconda3.sh
+#RUN conda --version
+
+## Install Mamba for faster installation
+#RUN conda install -c conda-forge mamba
+
+# Get Conda yml and install environment
+#RUN mamba env create -f /usr/local/TERA-Seq_manuscript/teraseq-env.yml
+RUN conda env create -f /root/TERA-Seq_manuscript/teraseq-env.yml
+
+# Increase default FastQC RAM
+RUN sed -i 's/-Xmx250m/-Xmx5g/g' /root/miniconda3/envs/teraseq/opt/fastqc-*/fastqc
+
+#ENV PATH="${PATH}:/root/miniconda3/envs/teraseq/bin"
+
+RUN ln -s /root/miniconda3/envs/teraseq/bin/R /bin/R \
+    && ln -s /root/miniconda3/envs/teraseq/bin/curl /bin/curl
+
+### Save default Conda path
+RUN sed -i '/CONDA_PREFIX/d' /root/TERA-Seq_manuscript/PARAMS.sh \
+    && echo -e "CONDA_PREFIX=\"/root/miniconda3\"" >> /root/TERA-Seq_manuscript/PARAMS.sh
+
+### Perl
+WORKDIR /root/TERA-Seq_manuscript/tools
+
+## System-wide install (option 1)
+# RUN cpanm inc::Module::Install \
+#     && cpanm autodie \
+#     && cpanm DBI \
+#     && cpanm Devel::Size \
+#     && cpanm Getopt::Long::Descriptive \
+#     && cpanm IO::File \
+#     && cpanm IO::Interactive \
+#     && cpanm IO::Uncompress::Gunzip \
+#     && cpanm Params::Validate \
+#     && cpanm Params::Util \
+#     && cpanm Sub::Install \
+#     && cpanm Modern::Perl \
+#     && cpanm --force MooseX::App::Simple \
+#     && cpanm --force MooseX::App::Command \
+#     && cpanm --force MooseX::Getopt::Meta::Attribute::Trait::NoGetopt
+#
+# WORKDIR /usr/local/share/perl/5.22.1
+#
+# RUN git clone --recursive https://github.com/genoo/GenOO.git GenOO_git
+# WORKDIR /usr/local/share/perl/5.22.1/GenOO_git
+# RUN git reset 6527029 --hard
+# WORKDIR /usr/local/share/perl/5.22.1
+# RUN mkdir GenOO \
+#     && cp -r GenOO_git/lib/GenOO/* GenOO/
+#
+# RUN cpanm CLIPSeqTools
+#
+# RUN wget https://raw.githubusercontent.com/mourelatos-lab/TERA-Seq_manuscript/main/misc/GenOOx/Data/File/SAMminimap2.pm -O /usr/local/share/perl/5.22.1/GenOOx/Data/File/SAMminimap2.pm
+# RUN mkdir /usr/local/share/perl/5.22.1/GenOOx/Data/File/SAMminimap2 \
+#     && wget https://raw.githubusercontent.com/mourelatos-lab/TERA-Seq_manuscript/main/misc/GenOOx/Data/File/SAMminimap2/Record.pm -O /usr/local/share/perl/5.22.1/GenOOx/Data/File/SAMminimap2/Record.pm
+
+## Virtual environment install (option 2)
+# Export Conda perl lib path (mainly for local::lib module)
+ENV PERL5LIB "/root/miniconda3/envs/teraseq/lib/site_perl/5.26.2/:${PERL5LIB}"
+ARG PERL5LIB="/root/miniconda3/envs/teraseq/lib/site_perl/5.26.2/:${PERL5LIB}"
+
+RUN git clone https://github.com/jizhang/perl-virtualenv.git \
+    && cd perl-virtualenv/ \
+    && git reset f931774 --hard \
+    && chmod u+x virtualenv.pl \
+    && ./virtualenv.pl teraseq \
+    && . teraseq/bin/activate \
+    && curl -L https://cpanmin.us/ -o teraseq/bin/cpanm \
+    && chmod +x teraseq/bin/cpanm
+
+RUN . perl-virtualenv/teraseq/bin/activate \
+    && cpanm inc::Module::[email protected] \
+    && cpanm [email protected] \
+    && cpanm [email protected] \
+    && cpanm Devel::[email protected] \
+    && cpanm Getopt::Long::[email protected] \
+    && cpanm IO::[email protected] \
+    && cpanm IO::[email protected] \
+    && cpanm --force IO::Uncompress::Gunzip \
+    && cpanm Params::[email protected] \
+    && cpanm Params::[email protected] \
+    && cpanm Sub::[email protected] \
+    && cpanm Modern::[email protected] \
+    && cpanm --force MooseX::App::[email protected] \
+    && cpanm --force MooseX::App::Command \
+    && cpanm --force MooseX::Getopt::Meta::Attribute::Trait::[email protected]
+
+RUN git clone --recursive https://github.com/genoo/GenOO.git perl-virtualenv/teraseq/lib/perl5/GenOO_git \
+    && cd perl-virtualenv/teraseq/lib/perl5/GenOO_git/ \
+    && git reset 6527029 --hard \
+    && cd ../ \
+    && mkdir GenOO \
+    && cp -r GenOO_git/lib/GenOO/* GenOO/
+
+# Install specific version of Perl module https://stackoverflow.com/questions/260593/how-can-i-install-a-specific-version-of-a-set-of-perl-modules
+RUN . perl-virtualenv/teraseq/bin/activate \
+    && cpanm --force [email protected]  \
+    && cp -r /root/TERA-Seq_manuscript/misc/GenOOx/* perl-virtualenv/teraseq/lib/perl5/GenOOx/
+
+################################################################################
+### Nanopolish
+# Default version
+RUN git clone --recursive https://github.com/jts/nanopolish.git \
+    && mv nanopolish nanopolish-480fc85 \
+    && cd nanopolish-480fc85/ \
+    && git reset 480fc85 --hard \
+    && sed -i 's#http://bitbucket.org/eigen/eigen/get/$(EIGEN_VERSION).tar.bz2#https://gitlab.com/libeigen/eigen/-/archive/$(EIGEN_VERSION)/eigen-$(EIGEN_VERSION).tar.bz2#' Makefile \
+    && sed -i 's/tar -xjf $(EIGEN_VERSION).tar.bz2/tar -xjf eigen-$(EIGEN_VERSION).tar.bz2/' Makefile \
+    && sed -i 's/eigen-eigen-\*/eigen-$(EIGEN_VERSION)/' Makefile \
+#    && sed -i '27 i EIGEN_VERSION_MV ?= d9c80169e091a2c6e75ceb509f81764d22cf6a63' Makefile \
+#    && sed -i 's/mv\ eigen-\$(EIGEN_VERSION)/mv\ eigen-\$(EIGEN_VERSION_MV)/' Makefile \
+    && rm -rf fast5 \
+    && git clone https://github.com/mateidavid/fast5.git \
+    && cd fast5/ \
+    && git reset 18d6e34 --hard \
+    && cd ../ \
+    && rm -rf htslib \
+    && git clone --recursive https://github.com/samtools/htslib.git \
+    && cd htslib/ \
+    && git reset 3dc96c5 --hard \
+    && cd ../ \
+    && make \
+    && ln -s $(pwd)/nanopolish /root/miniconda3/envs/teraseq/bin/nanopolish
+
+# New version with polya hmm scripts
+RUN git clone --recursive https://github.com/jts/nanopolish.git \
+    && mv nanopolish nanopolish-ab9722b \
+    && cd nanopolish-ab9722b/ \
+    && git reset ab9722b --hard
+
+################################################################################
+### Other dependencies
+# Make sure to activate Conda
+SHELL ["conda", "run", "-n", "teraseq", "/bin/bash", "-c"]
+
+## GeneCycle
+#RUN Rscript -e 'install.packages("GeneCycle", repos="https://cloud.r-project.org")'
+RUN Rscript -e 'install.packages(c("longitudinal", "fdrtool"), repos = "http://cran.us.r-project.org"); install.packages("https://cran.r-project.org/src/contrib/GeneCycle_1.1.5.tar.gz", repos=NULL, type="source")'
+
+## Cutadapt
+RUN mkdir cutadapt-2.5 \
+    && cd cutadapt-2.5/ \
+    && python3 -m venv venv \
+    && source venv/bin/activate \
+    && python3 -m pip install --upgrade pip \
+    && pip3 install cutadapt==2.5 pysam numpy pandas matplotlib seaborn \
+    && which cutadapt
+
+## DeepTools
+RUN mkdir deepTools-3.5.0 \
+    && cd deepTools-3.5.0/ \
+    && python3 -m venv venv \
+    && source venv/bin/activate \
+    && python3 -m pip install --upgrade pip \
+    && pip3 install wheel \
+    && pip3 install deeptools==3.5.0 \
+    && deeptools --version
+
+## ONT-Fast5-API
+RUN mkdir ont-fast5-api \
+    && cd ont-fast5-api/ \
+    && python3 -m venv venv \
+    && source venv/bin/activate \
+    && pip install ont-fast5-api==3.3.0 h5py seaborn
+
+## Jvarkit
+RUN git clone "https://github.com/lindenb/jvarkit.git" \
+    && mv jvarkit jvarkit-014d3e9 \
+    && cd jvarkit-014d3e9/ \
+    && git reset 014d3e9 --hard \
+    && ./gradlew biostar84452 \
+    && mkdir $CONDA_PREFIX/share/jvarkit \
+    && ln -s $(pwd)/dist/biostar84452.jar /root/miniconda3/envs/teraseq/share/jvarkit/remove-softlip.jar
+
+
+# Add utils dir to PATH
+ENV PATH "/usr/local/TERA-Seq_manuscript/tools/utils:${PATH}"
+
 WORKDIR /root/TERA-Seq_manuscript
-COPY . .
-RUN mv -f activate.d/* /root/miniconda3/envs/teraseq/etc/conda/activate.d
-RUN mv -f deactivate.d/* /root/miniconda3/envs/teraseq/etc/conda/deactivate.d
-WORKDIR /root/TERA-Seq_manuscript/data
-RUN ./run.sh
-WORKDIR /root/TERA-Seq_manuscript/samples

teraseq/misc/GenOOx/Data/File/SAMminimap2.pm

@@ -0,0 +1,65 @@
+# POD documentation - main docs before the code
+
+=head1 NAME
+
+GenOOx::Data::File::SAMminimap2 - GenOO framework extension to read SAM files created by the Minimap2 aligner
+
+=head1 SYNOPSIS
+
+GenOO framework extension to read SAM files created by the Minimap2 aligner.
+Include it in your script and ask GenOO SAM parser to use it.
+
+    use GenOOx::Data::File::SAMminimap2::Record;
+
+    my $file_parser = GenOO::Data::File::SAM->new(
+        file          => 'file.sam',
+        records_class => 'GenOOx::Data::File::SAMminimap2::Record'
+    );
+
+    while (my $record = $file_parser->next_record) {
+        # $record is now an instance of GenOOx::Data::File::SAMminimap2::Record.
+        print $record->cigar."\n"; # name
+        print $record->flag."\n"; # flag
+        print $record->number_of_mappings."\n"; # new stuff not present by default
+        print $record->best_hit."\n"; # new stuff not present by default
+        print $record->number_of_best_hits."\n"; # new stuff not present by default
+    }
+
+
+=head1 DESCRIPTION
+
+The GenOO framework SAM parser avoids code that is unique to specific programs and makes no assumptions for the optional fields in a SAM file. This module is a plugin for the GenOO framework and provides the functionality for reading SAM files generated from the Minimap2 aligner. The 
+module has been created on top of the generic GenOO SAM parser and to use it just include it in your scripts and ask GenOO SAM parser to use it.
+
+=head1 EXAMPLES
+
+    # Create a parser
+    my $file_parser = GenOO::Data::File::SAM->new(
+        file          => 'file.sam',
+        records_class => 'GenOOx::Data::File::SAMminimap2::Record'
+    );
+
+    # Loop on the records of the file
+    while (my $record = $file_parser->next_record) {
+        # $record is now an instance of GenOOx::Data::File::SAMminimap2::Record.
+        print $record->cigar."\n"; # name
+        print $record->flag."\n"; # flag
+        print $record->number_of_mappings."\n"; # new stuff not present by default in GenOO
+        print $record->best_hit."\n"; # new stuff not present by default in GenOO
+        print $record->number_of_best_hits."\n"; # new stuff not present by default in GenOO
+    }
+
+=cut
+
+# Let the code begin...
+
+package GenOOx::Data::File::SAMminimap2;
+$GenOOx::Data::File::SAMminimap2::VERSION = '0.0.1';
+
+#######################################################################
+#######################   Load External modules   #####################
+#######################################################################
+use Modern::Perl;
+
+
+1;

teraseq/misc/GenOOx/Data/File/SAMminimap2/Record.pm

@@ -0,0 +1,105 @@
+# POD documentation - main docs before the code
+
+=head1 NAME
+
+GenOOx::Data::File::SAMminimap2::Record - Represents a record of a SAM format file generated by Minimap2
+
+=head1 SYNOPSIS
+
+    # Object representing a record of a sam file 
+
+    # To initialize 
+    my $sam_record = GenOOx::Data::File::SAMminimap2::Record->new(
+        fields => [qname,flag, rname, pos, mapq, cigar,
+                   rnext, pnext, tlen, seq, qual, tags]
+    );
+
+
+=head1 DESCRIPTION
+
+    This object represents a record of a sam file generated by Minimap2 and offers methods for accessing the different
+    attributes. It implements several additional methods that transform original attributes in more manageable
+    attributes. eg. from the FLAG attribute the actual strand is extracted etc.
+
+=head1 EXAMPLES
+
+    # Check if the record corresponds to a match
+    my $mapped = $sam_record->is_mapped;
+
+    # Check if the record corresponds to a non match
+    my $unmapped = $sam_record->is_unmapped;
+
+    # Parse the FLAG attribute and return 1 or -1 for the strand
+    my $strand = $sam_record->strand;
+
+=cut
+
+# Let the code begin...
+
+package GenOOx::Data::File::SAMminimap2::Record;
+$GenOOx::Data::File::SAMminimap2::Record::VERSION = '0.0.1';
+
+#######################################################################
+#######################   Load External modules   #####################
+#######################################################################
+use Moose;
+use namespace::autoclean;
+
+
+#######################################################################
+############################   Inheritance   ##########################
+#######################################################################
+extends 'GenOO::Data::File::SAM::Record';
+
+
+#######################################################################
+########################   Interface Methods   ########################
+#######################################################################
+sub number_of_optimal_hits {
+	my ($self) = @_;
+
+	return $self->tag('X0:i') || 0;
+}
+
+sub number_of_suboptimal_hits {
+	my ($self) = @_;
+
+	return $self->tag('X1:i') || 0;
+}
+
+sub number_of_mappings {
+	my ($self) = @_;
+
+	return $self->number_of_optimal_hits + $self->number_of_suboptimal_hits;
+}
+
+sub best_hit {
+        my ($self) = @_;
+
+        return $self->tag('XP:i') || 0;
+}
+
+sub number_of_best_hits {
+        my ($self) = @_;
+
+        return $self->tag('XN:i') || 0;
+}
+
+sub alternative_mappings {
+	my ($self) = @_;
+
+	my @alternative_mappings;
+	my $value = $self->tag('XA:Z');
+	if (defined $value) {
+		@alternative_mappings = split(/;/,$value);
+	}
+	return @alternative_mappings;
+}
+
+
+#######################################################################
+############################   Finalize   #############################
+#######################################################################
+__PACKAGE__->meta->make_immutable;
+
+1;

teraseq/samples/.gitignore

@@ -0,0 +1 @@
+