Augustus Prediction

Augustus Prediction

1. Augustus Prediction - Run1

Augustus gene prediction run making use of all extrinsic evidence and weighted towards transcript assemblies along with RNA-Seq coverage

1.1 Start Augustus run: [~10 mins]

(
cd /home/train/Annotation_workshop/Augustus/Prediction/Chr3-1065466-1464870/run1
rm -rf Output && mkdir -p Output
/usr/bin/time -v augustus \
    --AUGUSTUS_CONFIG_PATH=../Inputs/config \
    --species=ath_wksp_opt_c32_k32 \
    --UTR=on \
    --extrinsicCfgFile=../Inputs/extrinsic.ei_augustus340_generic.cfg \
    --alternatives-from-evidence=true \
    --hintsfile=hints/extrinsic.augustus.run1.gff3 \
    --noInFrameStop=true \
    --allow_hinted_splicesites=atac \
    --outfile=Output/run1.raw.gtf \
    --errfile=Output/run1.log \
    --predictionStart=1065466 \
    --predictionEnd=1464870 \
    ../Inputs/Athaliana_447_TAIR10.Chr3.fa
)

1.2 Convert Augustus GTF into GFF3 format:

(
cd /home/train/Annotation_workshop/Augustus/Prediction/Chr3-1065466-1464870/run1
augustus_gtf_to_gff3 ARATH3702_run1 Output/run1.raw.gtf
)

2. Augustus Prediction - Run2

Augustus gene prediction run making use of all extrinsic evidence, same as run1, but without RNA-Seq coverage

2.1 Start Augustus run: [~3 mins]

(
cd /home/train/Annotation_workshop/Augustus/Prediction/Chr3-1065466-1464870/run2
rm -rf Output && mkdir -p Output
/usr/bin/time -v augustus \
    --AUGUSTUS_CONFIG_PATH=../Inputs/config \
    --species=ath_wksp_opt_c32_k32 \
    --UTR=on \
    --extrinsicCfgFile=../Inputs/extrinsic.ei_augustus340_generic.cfg \
    --alternatives-from-evidence=true \
    --hintsfile=hints/extrinsic.augustus.run2.gff3 \
    --noInFrameStop=true \
    --allow_hinted_splicesites=atac \
    --outfile=Output/run2.raw.gtf \
    --errfile=Output/run2.log \
    --predictionStart=1065466 \
    --predictionEnd=1464870 \
    ../Inputs/Athaliana_447_TAIR10.Chr3.fa
)

2.2 Convert Augustus GTF into GFF3 format:

(
cd /home/train/Annotation_workshop/Augustus/Prediction/Chr3-1065466-1464870/run2
augustus_gtf_to_gff3 ARATH3702_run2 Output/run2.raw.gtf
)

3. Augustus Prediction - Run3

Augustus gene prediction run makes use of all extrinsic evidence, same as run2, but weighted towards protein alignments

3.1 Start Augustus run: [~3 mins]

(
cd /home/train/Annotation_workshop/Augustus/Prediction/Chr3-1065466-1464870/run3
rm -rf Output && mkdir -p Output
/usr/bin/time -v augustus \
    --AUGUSTUS_CONFIG_PATH=../Inputs/config \
    --species=ath_wksp_opt_c32_k32 \
    --UTR=on \
    --extrinsicCfgFile=../Inputs/extrinsic.ei_augustus340_generic.cfg \
    --alternatives-from-evidence=true \
    --hintsfile=hints/extrinsic.augustus.run3.gff3 \
    --noInFrameStop=true \
    --allow_hinted_splicesites=atac \
    --outfile=Output/run3.raw.gtf \
    --errfile=Output/run3.log \
    --predictionStart=1065466 \
    --predictionEnd=1464870 \
    ../Inputs/Athaliana_447_TAIR10.Chr3.fa
)

3.2 Convert Augustus GTF into GFF3 format:

(
cd /home/train/Annotation_workshop/Augustus/Prediction/Chr3-1065466-1464870/run3
augustus_gtf_to_gff3 ARATH3702_run3 Output/run3.raw.gtf
)

4. Augustus Prediction - Run4

Ab initio Augustus gene prediction run without making use of any extrinsic evidence

4.1 Start Augustus run: [~1 mins]

(
cd /home/train/Annotation_workshop/Augustus/Prediction/Chr3-1065466-1464870/run4
rm -rf Output && mkdir -p Output
/usr/bin/time -v augustus \
    --AUGUSTUS_CONFIG_PATH=../Inputs/config \
    --species=ath_wksp_opt_c32_k32 \
    --UTR=on \
    --noInFrameStop=true \
    --allow_hinted_splicesites=atac \
    --outfile=Output/run4.raw.gtf \
    --errfile=Output/run4.log \
    --predictionStart=1065466 \
    --predictionEnd=1464870 \
    ../Inputs/Athaliana_447_TAIR10.Chr3.fa
)

4.2 Convert Augustus GTF into GFF3 format:

(
cd /home/train/Annotation_workshop/Augustus/Prediction/Chr3-1065466-1464870/run4
augustus_gtf_to_gff3 ARATH3702_run4 Output/run4.raw.gtf
)

5. Generate Mikado stats

5.1 Summarise the statistics generated by Mikado

(
for f in run*/*.stats.txt ; do
    tag=$(basename $f .txt)
    mkdir -p Mikado_stats
    parse_mikado_stats \
        ${f} \
        > Mikado_stats/${tag}.summary.txt
done
)

5.2 Concatenate all the annotation summary files

(
Scripts/paste_mikado_summary_stats.sh \
    Mikado_stats/*stats.summary.txt |
    tabulate -s "\t" \
    > Mikado_stats/all_gff.summary
)

5.3 Visualise the final output

cat Mikado_stats/all_gff.summary

6. Compare the Augustus models to the reference annotation

6.1 Use Mikado to compare the annotations to the reference annotation

(
for f in run*/ARATH3702*.gff; do
    tag=$(basename $f .gff)
    mikado compare \
        -r Inputs/Athaliana_447_Araport11.Chr3-1065466-1464870.gene_exons.gff3 \
        -p $f  \
        -o Mikado_Compare/${tag}
done
)

6.2 Summarise all Mikado comparisons

(
mikado util collect_compare \
    -fmt tsv \
    -l "all" \
    -o Mikado_Compare/Athaliana_augustus_models \
    Mikado_Compare/*.stats
)

6.3 Visualise the F1 metrics of all the Mikado comparisons

(
cat Mikado_Compare/Athaliana_augustus_models.f1.tsv \
    | sed -e 's:Mikado_Compare/::g' -e 's:.stats::g' \
    | tabulate -s "\t"
)