来自于跨国药企巨头的肺炎链球菌血清型预测工具

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]

# 下载程序代码
git clone https://github.com/PfizerRD/pfaster.git

# 安装依赖软件
conda env create -p /pub/apps/biotools/pfaster -f environment.yml

# 移动程序文件至 bin 目录
mv pfaster /pub/apps/biotools/pfaster/bin/pfaster

# 添加环境变量
PATH=/pub/apps/biotools/pfaster/bin:$PATH

# 切换目录
cd /pub/apps/biotools/pfaster/bin/pfaster

# 测试
python pfaster.py -f tests/Pn1_test_ERR1439829.fasta

/path/to/sample1.fasta
/path/to/sample2.fasta
/path/to/sample3.fasta
/path/to/sample4.fasta
/path/to/sample5.fasta
/path/to/sample6.fasta
/path/to/sample7.fasta
/path/to/sample8.fasta
/path/to/sample9.fasta
/path/to/sample10.fasta

PATH=/pub/apps/biotools/pfaster/bin:$PATH

OUTDIR=/path/to/results

echo -e "Sample_name\tIn_silico_serotype\tprobability" > pfaster_predict_results.tsv

whileread fasta
do
    cd /pub/apps/biotools/pfaster/bin/pfaster
    python pfaster.py -f $fasta -o $OUTDIR

    seq=$(sed -n '1 p'$OUTDIR/prediction.txt | sed 's#.*/##')
    st=$(sed -n '2 p'$OUTDIR/prediction.txt)
    prob=$(sed -n '3 p'$OUTDIR/prediction.txt)
    cd -
    echo -e "$seq\t$st\t$prob" >> pfaster_predict_results.tsv
done < lst

Sample_name    In_silico_serotype    probability
sample1    3    0.89
sample2    15C    0.67
sample3    6E    1.0
sample4    19A    1.0
sample5    19A    1.0
sample6    19A    1.0
sample7    14    0.99
sample8    3    1.0
sample9    19F    1.0
sample10    19F    0.99

cd /pub/apps/biotools/pfaster/bin/pfaster