Skip to content

Latest commit

 

History

History
106 lines (71 loc) · 3.59 KB

evidence-based.md

File metadata and controls

106 lines (71 loc) · 3.59 KB
Raw

Gene Prediction

1. Evidence Based Gene Prediction

Overview of steps:

The steps for evidence based prediction are as follows:

  1. Map RNAseq reads to the genome
  2. Assemble transcripts using various transcript assemblers
  3. Run Mikado to consolidate transcripts and run programs to identify splice junctions, ORFs and full length matches to SwissProt plants proteins.
  4. Pick transcripts for each locus and finalize evidence based annotations.
  5. Post-processing of evidence based predictions.

Steps in detail:

1. Map RNAseq reads:

Reads were mapped using STAR mapping program as follows:

# index
runSTARmap_index.sh NAM.pseudomolecules.fasta
# first round mapping
for read1 in *R1.fq.gz; do
  runSTARmap_round1.sh NAM.pseudomolecules.fasta ${read1}
done
# consolidate splice info
awk -f sjCollapseSamples.awk *_SJ.out.tab | sort -k1,1V -k2,2n -k3,3n > SJ.all
# second round mapping
for read1 in *R1.fq.gz; do
  runSTARmap_round2.sh NAM.pseudomolecules.fasta ${read1}
done
# merge BAM files
merge-bam-files.sh NAM

2. Transcript Assembly

Five different transcript assemblers were ran on the merged BAM file (merged_NAM.bam) as follows:

runStrawberry.sh merged_NAM.bam
runStringtie.sh merged_NAM.bam
runTrinity.sh merged_NAM.bam
runClass2.sh merged_NAM.bam
runCufflinks.sh merged_NAM.bam

Since Trinity generates a fasta file, we will use GMAP to map them and create a GFF3 file.

runGMAPcdna.sh NAM.pseudomolecules.fasta

We will also need splice junctions, that we can generate using PortCullis:

runPortcullis.sh merged_NAM.bam

3. Consolidate transcripts and prepare files

Here, we will use all the transcript assemblies generated in the previous step, along with splice junctions and pool them together. The redundant transcripts are also removed. Using consolidated transcripts, we will find full length trasncritps by BLAST searching against SwissProt proteins, and detect ORFs using TransDecoder.

This is all accomplished using this script:

copy-files-for-mikado.sh NAM

4. Finalize evidence-based predictions

As a last step, we will run Mikado to finalize the predictions. It is done using this script:

finalize-files-and-pick.sh NAM

5. Post processing of evidence-based predictions

Additional structural improvements for the Mikado generated transcripts were completed using the PASA (v2.3.3) genome annotation tool. The inputs for PASA included 2,019,896 maize EST derived from genbank, linage specific Mikado transcripts, 69,163 B73 full length cDNA from genbank and 46,311 maize iso-seq transcripts from 11 developmental tissues that were filtered for intron retention. Details on how the evidence for PASA was curated is provided here:?? PASA updated the models, providing UTR extensions, novel and additional alternative isoforms.

PASA was run in there steps:

Step 1. included aligning the evidence to the repeatmasked genome sequence. All the evidences were combined into a single fasta input, along with masked genome, pasa-config file and Full-length accession list to the script pasa_run_1.sh

qsub pasa_run_1.sh

Step 2. included loading the pasa transcript assemblies to sqlite database for comparsion and updating the Mikado models. This was done using script pasa_run_2.sh

qsub pasa_run_2.sh

Step 3. Lastly PASA was again run updating models from Step 2 usng the script pasa_run_3.sh

qsub pasa_run_2.sh