Merge pull request #1047 from taylorpaisie/tkp-seqfu

adds SeqFu

Program_Licenses.md

@@ -155,6 +155,7 @@ The licenses of the open-source software that is contained in these Docker image
 | SalmID| MIT | https://github.com/hcdenbakker/SalmID/blob/master/LICENSE |
 | samclip | GNU GPLv3 | https://github.com/tseemann/samclip/blob/master/LICENSE |
 | Samtools | GNU GPLv3 | https://github.com/samtools/samtools/blob/develop/LICENSE |
+| SeqFu | GNU GPLv3 | https://github.com/telatin/seqfu2 |
 | SeqKit | MIT | https://github.com/shenwei356/seqkit/blob/master/LICENSE
 | SeqSero | GNU GPLv2 | https://github.com/denglab/SeqSero/blob/master/LICENSE |
 | SeqSero2 | GNU GPLv2 | https://github.com/denglab/SeqSero2/blob/master/LICENSE |

README.md

@@ -266,6 +266,7 @@ To learn more about the docker pull rate limits and the open source software pro
 | [SalmID](https://hub.docker.com/r/staphb/salmid) <br/> [![docker pulls](https://badgen.net/docker/pulls/staphb/salmid)](https://hub.docker.com/r/staphb/salmid) | <ul><li>0.1.23</li></ul> | https://github.com/hcdenbakker/SalmID |
 | [samclip](https://hub.docker.com/r/staphb/samclip) <br/> [![docker pulls](https://badgen.net/docker/pulls/staphb/samclip)](https://hub.docker.com/r/staphb/samclip) | <ul><li>[0.4.0](./samclip/0.4.0/)</li></ul>| https://github.com/tseemann/samclip |
 | [Samtools](https://hub.docker.com/r/staphb/samtools) <br/> [![docker pulls](https://badgen.net/docker/pulls/staphb/samtools)](https://hub.docker.com/r/staphb/samtools) | <ul><li>[1.9](./samtools/1.9/)</li><li>[1.10](./samtools/1.10/)</li><li>[1.11](./samtools/1.11/)</li><li>[1.12](./samtools/1.12/)</li><li>[1.13](./samtools/1.13/)</li><li>[1.14](./samtools/1.14/)</li><li>[1.15](./samtools/1.15/)</li><li>[1.16](./samtools/1.16/)</li><li>[1.16.1](./samtools/1.16.1/)</li><li>[1.17](./samtools/1.17/)</li><li>[1.17-2023-06](./samtools/1.17-2023-06/)</li><li>[1.18](./samtools/1.18/)</li><li>[1.19](./samtools/1.19/)</li><li>[1.20](./samtools/1.20/)</li><li>[1.20.c](./samtools/1.20.c/)</li></ul> | https://github.com/samtools/samtools |
+| [SeqFu](https://hub.docker.com/r/staphb/SeqFu) <br/> [![docker pulls](https://badgen.net/docker/pulls/staphb/seqfu)](https://hub.docker.com/r/staphb/seqfu) | <ul><li>[1.20.3](./seqfu/1.20.3/)</li></ul> |  https://github.com/telatin/seqfu2 |
 | [SeqKit](https://hub.docker.com/r/staphb/SeqKit) <br/> [![docker pulls](https://badgen.net/docker/pulls/staphb/seqkit)](https://hub.docker.com/r/staphb/seqkit) | <ul><li>[2.3.1](./seqkit/2.3.1/)</li><li>[2.6.1](./seqkit/2.6.1/)</li><li>[2.7.0](./seqkit/2.7.0/)</li><li>[2.8.0](./seqkit/2.8.0/)</li><li>[2.8.1](./seqkit/2.8.1/)</li><li>[2.8.2](./seqkit/2.8.2/)</li></ul> | https://github.com/shenwei356/seqkit |
 | [SeqSero](https://hub.docker.com/r/staphb/seqsero/) <br/> [![docker pulls](https://badgen.net/docker/pulls/staphb/seqsero)](https://hub.docker.com/r/staphb/seqsero) | <ul><li>1.0.1</li></ul> | https://github.com/denglab/SeqSero |
 | [SeqSero2](https://hub.docker.com/r/staphb/seqsero2/) <br/> [![docker pulls](https://badgen.net/docker/pulls/staphb/seqsero2)](https://hub.docker.com/r/staphb/seqsero2) | <ul><li>[0.1.0](./seqsero2/0.1.0/)</li><li>[1.0.0](./seqsero2/1.0.0/)</li><li>[1.0.2](./seqsero2/1.0.2/)</li><li>[1.1.0](./seqsero2/1.1.0/)</li><li>[1.1.1](./seqsero2/1.1.1/)</li><li>[1.2.1](./seqsero2/1.2.1/)</li><li>[1.3.1](./seqsero2/1.3.1/)</li></ul> | https://github.com/denglab/SeqSero2/ |

seqfu/1.20.3/Dockerfile

@@ -0,0 +1,90 @@
+# Set global variables
+ARG SEQFU_VER="1.20.3"
+ARG NIM_VER="1.22.2"
+
+# Stage 1: Build Dockerfile
+FROM ubuntu:focal AS builder
+ARG SEQFU_VER
+
+# Set non-interactive frontend and update PATH
+ENV DEBIAN_FRONTEND=noninteractive \
+    PATH="/root/.nimble/bin:${PATH}"
+
+# Install required dependencies
+RUN apt-get update && apt-get install -y --no-install-recommends \
+    wget \
+    unzip \
+    build-essential \
+    zlib1g-dev \
+    curl \
+    python3 \
+    python3-pip \
+    git \
+    ca-certificates && \
+    rm -rf /var/lib/apt/lists/*
+
+# Set the working directory
+WORKDIR /root
+
+# Install Nim and SeqFu
+RUN curl https://nim-lang.org/choosenim/init.sh -sSf | sh -s -- -y --ver ${NIM_VER}
+
+# Install SeqFu
+RUN wget https://github.com/telatin/seqfu2/archive/refs/tags/v${SEQFU_VER}.tar.gz && \
+    tar -xzf v${SEQFU_VER}.tar.gz && \
+    rm v${SEQFU_VER}.tar.gz && \
+    cd seqfu2-${SEQFU_VER} && \
+    make
+
+# Stage 2: Create the final image
+FROM ubuntu:focal AS app
+ARG SEQFU_VER
+
+# Metadata
+LABEL base.image="ubuntu:focal" 
+LABEL dockerfile.version="1" 
+LABEL software="SeqFu"
+LABEL software.version="${SEQFU_VER}"
+LABEL description="A general-purpose program to manipulate and parse information from FASTA/FASTQ files, supporting gzipped input files."
+LABEL website="https://github.com/telatin/seqfu2"
+LABEL license.url="https://github.com/telatin/seqfu2?tab=GPL-3.0-1-ov-file#readme"
+LABEL maintainer="Taylor K. Paisie"
+LABEL maintainer.email="[email protected]"
+
+# Copy the necessary files from the builder
+COPY --from=builder /root/seqfu2-${SEQFU_VER}/bin /usr/local/bin
+
+# Install minimal runtime dependencies
+# curl is used in checking for the versions of dependencies
+RUN apt-get update && apt-get install -y --no-install-recommends \
+    curl \
+    ca-certificates \
+    wget && \
+    apt-get autoclean && rm -rf /var/lib/apt/lists/*
+
+ENV LC_ALL=C
+
+CMD seqfu --help
+
+# Stage 3: testing the installation
+FROM app AS test
+
+ARG SEQFU_VER
+
+# Verify installation
+RUN seqfu --version && seqfu --help
+
+WORKDIR /test
+
+# using dev supplied tests
+COPY --from=builder /root/seqfu2-${SEQFU_VER} /test
+
+# skipping test-check.sh because it throws an error on dev-supplied fastq files
+RUN mv test/test-check.sh test/test-check.sh_skip && \
+    bash test/mini.sh
+
+# testing on user-supplied files
+RUN wget -q https://github.com/StaPH-B/docker-builds/raw/master/tests/SARS-CoV-2/SRR13957123_1.fastq.gz && \
+    wget -q https://github.com/StaPH-B/docker-builds/raw/master/tests/SARS-CoV-2/SRR13957123_2.fastq.gz && \
+    seqfu check SRR13957123_1.fastq.gz SRR13957123_2.fastq.gz > seqfu_check_test.txt && \
+    seqfu count -f SRR13957123_1.fastq.gz -r SRR13957123_2.fastq.gz > seqfu_count_test.txt

seqfu/1.20.3/README.md

@@ -0,0 +1,52 @@
+# SeqFu
+
+Main tool: [SeqFu](https://github.com/telatin/seqfu2)
+
+Code repository: https://github.com/telatin/seqfu2
+
+Basic information on how to use this tool:
+- executable: |
+```
+  · count [cnt]         : count FASTA/FASTQ reads, pair-end aware
+  · deinterleave [dei]  : deinterleave FASTQ
+  · derep [der]         : feature-rich dereplication of FASTA/FASTQ files
+  · interleave [ilv]    : interleave FASTQ pair ends
+  · lanes [mrl]         : merge Illumina lanes
+  · list [lst]          : print sequences from a list of names
+  · metadata [met]      : print a table of FASTQ reads (mapping files)
+  · sort [srt]          : sort sequences by size (uniques)
+  · stats [st]          : statistics on sequence lengths
+
+  · cat                 : concatenate FASTA/FASTQ files
+  · grep                : select sequences with patterns
+  · head                : print first sequences
+  · rc                  : reverse complement strings or files
+  · tab                 : tabulate reads to TSV (and viceversa)
+  · tail                : view last sequences
+  · view                : view sequences with colored quality and oligo matches
+```
+
+- help: `seqfu --help`
+- version: `seqfu --version`
+- description: | 
+> A general-purpose program to manipulate and parse information from FASTA/FASTQ files, supporting gzipped input files. 
+
+
+Full documentation: https://telatin.github.io/seqfu2/
+
+
+# Testing SeqFU analysis
+```
+wget -q https://github.com/StaPH-B/docker-builds/raw/master/tests/SARS-CoV-2/SRR13957123_1.fastq.gz
+
+wget -q https://github.com/StaPH-B/docker-builds/raw/master/tests/SARS-CoV-2/SRR13957123_2.fastq.gz
+
+seqfu check \
+  SRR13957123_1.fastq.gz \
+  SRR13957123_2.fastq.gz
+
+seqfu count \
+  -f SRR13957123_1.fastq.gz \
+  -r SRR13957123_2.fastq.gz
+```
+