tomoakin/Jin-annot

Scripts for annotating a genome with RNA-seq and Protein sequence data.

Jin-annot

Scripts for annotating a genome with RNA-seq and Protein sequence data.

This repository holds scripts and programs that was used to annotate a plant genome using RNA-seq and annotation of other sequenced genomes.

The scripts are making it easy to execute what are desribed at various pages. Important pages are:

• http://bioinf.uni-greifswald.de/bioinf/wiki/pmwiki.php?n=IncorporatingRNAseq.Tophat
• http://bioinf.uni-greifswald.de/bioinf/wiki/pmwiki.php?n=Augustus.Augustus
• http://www.repeatmasker.org/

Most step are encupsulated in job scripts that can be run in a SGE/UGE environment, while some are left for command line execution. The job scripts were run on the phase 1 system in NIG super computer. Some parameters, especially job parameters regarding parallel environment need adjustment to the system you use.

Most of the software used in these scripts are installed with LPM.

• http://www.kasahara.ws/lpm/

Input files

• Jin03.fa (the genome not included)
• RNA-seq-list (the list of RNA-seq data; index number and description) The raw RNA-seq data are organized in subdirectories as by index numbers

Sample_idx1/
  idx1_ATCACG_L006_R1_001.fastq.gz
  idx1_ATCACG_L006_R2_001.fastq.gz

RepeatMask

Masking repeats is recommended by Augustus author. To mask repeat properly, we need a proper species specific repeat profile.

Repeat Modeling

• Jin3RM.job
• Jin3RM2.job

Actual Repeat Masking

• repeatmask2.job

Note the directory name RM_65194.FriMay90932462014 should be changed to what was made by repeat modeler.

CEGMA

• cegma03.job

Interleave read1 and read2 of RNA-seq while trimming the adapter sequence

• cutadapt-pe.c

Rename the readname from /1,/2 to -1,-2

• rename-aug.job

Iterative Augustus

Initial training of Augustus with CEGMA results

new_species.pl --species=Jinfusca3   

• etrain.job

We need to train augustus with some data. At begining just use the ultra conserved genes from cegma.

The instruction was given at:

• http://bioinf.uni-greifswald.de/bioinf/wiki/pmwiki.php?n=Augustus.CEGMATraining

Prepare Bowtie2 index

• bb.job

Because tophat2 use bowtie2, bowtie2 index should prepared in advance. This is run as a single process, while actuall mapping process can be run in parallel, using the same index and separate query RNA-seq data. While this produces index files starting Jin03_m_db, tophat process requires presence of Jin03_m_db.fa

ln Jin03.masked.fa Jin03_m_db.fa

should be performed.

Map the RNA-seq data to the masked genome with tophat

• tophat2masked.job

• filter.job

Select good mapping reads. (--paired for paired end read data) This phase is run independently for each index of RNA-seq data.

• merge.job

Merge the filtered data and sort. Because we merge, we have to wait all filter job finished and run as a separte single job.

Generate hints file from the mapped data

• genhint.job

Split genome

• flatsplitbysize.rb

mkdir -p split
cd split
cp -p ../Jin03_m_db.fa .
ruby ../flatsplitbysize.rb Jin03_m_db.fa 1000000

Split the genome fasta file in a chunk of 1 Mb. When the scaffold is large one scaffold is in the split file, while multiple scaffold is put in a single file until the cumulative size reach 1 Mb.

Perform 1st augustus run

• augustus1.job
• split/augsp.job

Initially, augustus1.job was used. This took a day or more. After realizing that the genome may split and augustus is run in parallel split/augsp.job was written.

Extract exon junctions

• create_exex_junction.sh
• s7create_exex_junction.sh

Step 7

Map the RNA-seq to exon junctions using bowtie2

• bowtie2exexj.job
• s8bowtie2exexj.job

Map to exon junction database and adjust the coordinate back to the global coordinate.

• discard_intron_map.job

_thm/accepted_hits.noN.bam

merge2.job

Discard the intron map data from original mapping.

• s9join.job
• s10filter.job
• s12augustus2.job
• s13joingff.job
• s14togbandretrain.job

Map protein sequence data to genome using exonerate

• exonerateAt.job
• exoneratePp.job

Each of these jobs does map the protein sequences to the masked draft genome.

• exonerate-merge-to-hints.job

collects all the output file of the exonerateAt.job and exoneratePp.job and converts to hint file. The output files are identified by the name exonerate??.job.o*.

Convert repeatmasker output to hints.

• rm2hints.job

Construct combined hint data.

• mergehints.job

Run Augustus on split genome

• s22baugustus3.job

• s23bjoingff.job

Combine the split prediction over the genome.

s22augustus3.job
s23joingff.job

These output in gff2 format whereas the above is configured to write gff3 format.

unexplained

* exclude_intron_map.sh
s15optimize_augustus.job