Skip to content

Latest commit

 

History

History
215 lines (121 loc) · 5.95 KB

Gene-prediction.md

File metadata and controls

215 lines (121 loc) · 5.95 KB
Raw

Tetep gene prediction with MAKER-P (http://weatherby.genetics.utah.edu/MAKER/wiki/index.php/MAKER_Tutorial)


1) Pre-process procedures

1.1) Get homologous EST and protein data from NCBI (https://www.ncbi.nlm.nih.gov/nucest) and Uniprot (https://www.uniprot.org/), format the downloaded data

perl -ne 'next if (/^\s+$/); s/\>gi\|(\d+).*/>$1/; print;' \
    est/Oryza_sativa.est.NCBI-20160406.fasta \
    > est/Oryza_sativa.est.NCBI-20160406.rename.fasta

perl -ne 'next if (/^\s+$/); s/\>sp\|(\w+).*/>$1/; print;' \
    UniProt/uniprot-Oryza_sativa-20160527.fasta \
    > UniProt/uniprot-Oryza_sativa-20160527.rename.fasta

1.2) Process RNA-seq data

Step1: align RNA reads to Tetep genome using STAR aligner (https://github.com/alexdobin/STAR)

STAR --genomeDir star \
    --readFilesIn Tetep_RNA_1.fq.gz Tetep_RNA_1_2.fq.gz \
    --runThreadN 12 --readFilesCommand zcat \
    --outSAMstrandField intronMotif --outFilterIntronMotifs RemoveNoncanonical \
    --outSAMtype BAM SortedByCoordinate \
    --outFileNamePrefix Tetep.star

samtools index Tetep.starAligned.sortedByCoord.out.bam

Step2: Mapping transcriptome to reference genome with Cufflinks (http://cole-trapnell-lab.github.io/cufflinks/cufflinks/index.html)

cufflinks Tetep.starAligned.sortedByCoord.out.bam \
    –frag-bias-correct genome.fasta \
    –multi-read-correct –max-intron-length 20000 –no-update-check

Step3: Transcriptome denovo with Trinity (https://github.com/trinityrnaseq/trinityrnaseq/wiki)

Step3-1: Genome-free assembly

Trinity --CPU 12 --seqType fq --max_memory 150G --full_cleanup \
    --samples_file trinity/Tetep_RNA.samples.txt \
    --output trinity/Tetep.star.trinity

Step3-2: Genome-guided assembly

Trinity --CPU 6 --max_memory 20G --genome_guided_max_intron 10000 \
    --genome_guided_bam Tetep.starAligned.sortedByCoord.out.bam \
    --output trinity/Tetep.star.trinity

Step3-3: combine results from both assembly

cat trinity/Tetep.star.trinity.fasta \
    trinity/Tetep.star.trinity-GG.fasta \
    > trinity/Tetep.star.trinity-all.fasta

1.3) assessing genome assembly and annotation completeness with CEGMA (Core Eukaryotic Genes Mapping Approach, https://github.com/KorfLab/CEGMA_v2/)

cegma -T 8 -g genome.fasta -o genome


2) MAKER-P pipeline

Step1: Train SNAP (https://github.com/KorfLab/SNAP) with CEGMA generated gff output

cd snap/cegma/

cegma2zff genome.cegma.gff genome.fasta

fathom -categorize 1000 genome.ann genome.dna

fathom -export 1000 -plus uni.ann uni.dna

forge export.ann export.dna

hmm-assembler.pl genome.cegma.snap . genome.cegma.snap.hmm

Step2: First-pass run of MAKER, use:

  1. EST data (trinity assembled mRNAs, data downloaded from NCBI dbEST, keyword: Oryza sativa, retrieved: 20160406)

    est=trinity/Tetep.star.trinity-all.fasta

    altest=est/Oryza_sativa.est.NCBI-20160406.rename.fasta

    est2genome=1

  2. Proteins (data downloaded from UniProt, keyword: Oryza sativa, retrieved: 20160527)

    protein=UniProt/uniprot-Oryza_sativa-20160527.rename.fasta

    protein2genome=1

  3. a trained SNAP model

    snaphmm=snap/genome.cegma.snap.hmm

  4. predicted gff3 file from cufflinks

    pred_gff=star/Tetep.starAligned.sortedByCoord.cufflinks.transcripts.gff3

  5. other settings

    min_contig=10000

    min_protein=30

    alt_splice=1

mpiexec -mca btl ^openib -n 12 maker \
    -genome genome.fasta \
    -cpus 1 -quiet -fix_nucleotides -base Tetep

Step3: Generate GFF3 file from 1st pass run

gff3_merge -n -d maker/pass1/Tetep.maker.output/master_datastore_index.log \
    -o maker/pass1/genome.maker.pass1.gff

Step4: Further training SNAP and Augustus (http://augustus.gobics.de/) use 1st pass results

Step4-1: Training Augustus

autoAug.pl --genome=genome.fasta \
    --species=oryza_sativa_indica --singleCPU --useexisting \
    --trainingset=maker/pass1/genome.maker.pass1.gff \
    --workingdir=augustus/maker

Step4-2: Re-generate SNAP traning file

maker2zff -n -d maker/pass1/Tetep.maker.output/master_datastore_index.log

fathom -categorize 1000 snap/maker/genome.ann \
    snap/maker/genome.dna

fathom -export 1000 -plus snap/maker/uni.ann \
    snap/maker/uni.dna

forge snap/maker/export.ann \
    snap/maker/export.dna

hmm-assembler.pl genome.maker.snap . \
    > snap/genome.maker.snap.hmm

Step5: Second-pass run of MAKER, using trained Augustus and SNAP model

  1. EST data (trinity assembled mRNAs, data downloaded from NCBI dbEST, keyword: Oryza sativa, retrieved: 20160406)

    est=trinity/Tetep.star.trinity-all.fasta

    altest=est/Oryza_sativa.est.NCBI-20160406.rename.fasta

    est2genome=0

  2. Proteins (data downloaded from UniProt, keyword: Oryza sativa, retrieved: 20160527)

    protein=UniProt/uniprot-Oryza_sativa-20160527.rename.fasta

    protein2genome=0

  3. a re-trained SNAP model

    snaphmm=snap/genome.maker.snap.hmm

  4. a trained Augustus model

    augustus_species=oryza_sativa_indica

  5. predicted gff3 file from cufflinks

    pred_gff=star/Tetep.starAligned.sortedByCoord.cufflinks.transcripts.gff3

  6. other settings

    min_contig=10000

    min_protein=30

    alt_splice=1

mpiexec -mca btl ^openib -n 4 maker \
    -genome genome.fasta \
    -cpus 1 -quiet -fix_nucleotides -base Tetep

Step6: Generate new GFF3 annotation file

gff3_merge -n -g -d maker/pass2/Tetep.maker.output/master_datastore_index.log \
    -o maker/pass2/genome.maker.pass2.gff3


fasta_merge -d maker/pass2/Tetep.maker.output/master_datastore_index.log \
    -o maker/pass2/genome.maker.pass2