本攻略最早更新在https://github.com/gotouerina/Annotation，本文为将自己写的大粪再搬到简书上。本文需要的脚本都在git上有需要可以自取。

RepeatElements Annotation

01. Repeatmoderer

BuildDatabase  -name  $name  $fasta
Repeatmodeler -pa 20 -database $name    -LTRStruct

02.DeepTE

grep "ltr" consensi.fa.classified |grep "Unknown" > ltr_unknown.list
grep "rnd" consensi.fa.classified |grep "Unknown" > rnd_unknown.list
grep '>' Parascalop-families.fa  |  grep -v 'Unknown' > known.list
perl extract.pl ltr_unknown.list $family.fa > ltr_unknown.fa
perl extract.pl rnd_unknown.list $family.fa > rnd_unknown.fa
perl extract.pl known.list $family.fa > known.fa

Then

mkdir Un_ltr
mkdir Un_rnd
python3 /data/01/p1/user192/software/DeepTE/DeepTE.py -d working_ltr -o Un_ltr -i ltr_unknown.fa -sp M -m_dir /data/01/user214/RepeatWork/10.annotation/Parascalops/Metazoans_model  -fam LTR
python3 /data/01/p1/user192/software/DeepTE/DeepTE.py -d working_rnd -o Un_rnd -i rnd_unknown.fa -sp M -m_dir /data/01/user214/RepeatWork/10.annotation/Parascalops/Metazoans_model 

03.RepeatMasker

RepeatMasker Parascalops.v3.fa  -lib recalss.families.fa -e rmblast -xsmall -s -gff -pa 20

Denovo Prediction

01.Augustus

    mkdir fa-split; cd fa-split
    ln -s /path/to/$soft_mask
    perl split.pl $soft_mask && rm $soft_mask
    for i in fa-split/*.fa; do echo -e "/data/00/software/augustus/augustus-3.3.3/bin/augustus --softmasking=1 --species=human  $i > $i.out" >> augustus.sh; done #submit to slurm
    for i in fa-split/*.out; do echo -e "perl /data/01/user157/utils/ConvertFormat_augustus.pl $i  $i.gff" >> augustus2.sh ; done #submit to slurm
    cat fa-split/*.gff > augustus.gff3

02.GlimmerHmm

Glimmerhmm主要用于植物，预测的基因数量很多，外显子小，如果对动物基因组做注释，此步可以不做。。

    for i in fa-split/*.fa; do echo -e "/data/00/software/GlimmerHMM/GlimmerHMM/bin/glimmerhmm_linux_x86_64   $i  -d /data/00/software/Gl2/software/genscan/HumanIso.simmerHMM/GlimmerHMM/trained_dir/human    -g  -o $i.gff" >> glimmerhmm.sh; done
    cat fa-split/*.gff > glimmerhmm.gff3
    perl /data/01/user203/project/guohuai/03.gene_predict/utils/ConvertFormat_glimmerhmm.pl  glimmerhmm.gff3 glimmerhmm.final.gff3

03.GenScan

    perl split_genescan.pl $soft_mask
    for i in genscan_temp/*.fa; do echo -e " /data/00/user/user112/software/genscan/genscan /data/00/user/user112/software/genscan/HumanIso.smat   $i > $i.genescan "  >> genscan.sh; done #submit to slurm
    for i in genscan_temp/*.genescan; do echo -e "perl  /data/01/user194/yxj/shutu/yxj/01annotion/jiaoben/ConvertFormat_genscan.pl $i > $i.gff" >> genscan2.sh; done #submit to slurm
    cat genscan_temp/*.gff > genescan.raw.gff3
    cat genescan.raw.gff3 | perl -F'\t' -alne '$_ =~ m/(-.*)\s+genscan/; s/$1//g; print' >  genescan.final.gff3 

Homology Prediction

     source /data/00/user/user157/miniconda3/bin/activate ; conda activate GeMoMa
     java -jar -Xmx80g /data/00/user/user157/miniconda3/envs/GeMoMa/share/gemoma-1.7.1-0/GeMoMa-1.7.1.jar CLI GeMoMaPipeline threads=40  t=$soft_mask   s=own  g=$ref.fasta  a=$ref.prefix   outdir=$out_dir     AnnotationFinalizer.r=NO tblastn=false  ## do this for at least three species!
     cd  $out_dir
     perl /data/01/user203/project/guohuai/03.gene_predict/utils/ConvertFormat_GeMoMa.pl  final_annotation.gff

Transcript Prediction

    第一步对基因组建索引，$soft_mask是基因组文件，$index自定义名字，一般是物种名
    /data/00/software/hisat/hisat2-2.1.0/hisat2-build $soft_mask $index
    第二步比对，如果有多个样本，分别比对然后samtools merge到一起
    /data/00/software/hisat/hisat2-2.1.0/hisat2 --dta -p 20 -x $index -1 $R1.fq.gz  -2 $R2.fq.gz  | /data/00/software/samtools/samtools-1.15.1/samtools  sort -@ 10 >  trans.bam
    第三步，组装转录本
    /data/00/software/stringtie/stringtie-2.2.1/stringtie -p 10 -o merged.gtf trans.bam
    第四步，transdecoder做基因预测（此部分属于other prediction）
    /data/00/software/TransDecoder/TransDecoder-TransDecoder-v5.5.0/util/gtf_to_alignment_gff3.pl merged.gtf > transcripts.gff3
    /data/00/software/TransDecoder/TransDecoder-TransDecoder-v5.5.0/util/gtf_genome_to_cdna_fasta.pl merged.gtf  $soft_mask  > transcripts.fasta
    /data/00/software/TransDecoder/TransDecoder-TransDecoder-v5.5.0/TransDecoder.LongOrfs -t transcripts.fasta
    /data/00/software/TransDecoder/TransDecoder-TransDecoder-v5.5.0/TransDecoder.Predict -t trafnscripts.fasta
    /data/00/software/TransDecoder/TransDecoder-TransDecoder-v5.5.0/util/cdna_alignment_orf_to_genome_orf.pl transcripts.fasta.transdecoder.gff3  transcripts.gff3   transcripts.fasta  > transcripts.fasta.transdecoder.genome.gff3
    第五步，做pasapipeline
    $PASAHOME/bin/seqclean  transcripts.fasta
    $PASAHOME/Launch_PASA_pipeline.pl    -c alignAssembly.config -C -R -g genome.fasta            -t all_transcripts.fasta.clean -T -u all_transcripts.fasta      --ALIGNERS blat,gmap --CPU 10
   sample_mydb_pasa.pasa_assemblies.gff3是最终结果

如果用容器的话
$PASAHOME/改成 singularity pasapipeline.v2.5.3.simg /usr/local/src/PASApipeline

EvidenceModerler Merge

    cat transcripts.fasta.transdecoder.genome.gff3 genescan.final.gff3  glimmerhmm.final.gff3 augustus.gff3 > denovo.gff3
    cat $out_dir1/final_annotation.gff.for.evm  $out_dir2/final_annotation.gff.for.evm $out_dir3/final_annotation.gff.for.evm > homology.gff3
    /data/01/user214/RepeatWork/10.annotation/GSmole/08.EVM/EVidenceModeler-v2.1.0/EVidenceModeler  --genome $mask_soft.fasta  --sample_id $prefix  --gene_predictions denovo.gff3 --protein_alignments homology.gff3    --transcript_alignments transcripts.gff3    --segmentSize 100000   --overlapSize 10000 --cpu 20 --weights weights.txt