Alignment to diploid genome using STAR

[Bishwa Kiran]

Hi Alex,

I just wanted to follow upon a little bit more on ASE (allele specific expression)  using diploid genome. We had a brief discussion about it sometime ago.

https://groups.google.com/forum/#!searchin/rna-star/diploid$20genome|sort:relevance/rna-star/zs7qsaDidcA/acKagxYfBAAJ

So, now I am able to create 1) dipoid genome by inserting the phased SNPs and InDels in the haploid reference 2) GTF/GFF file with updated GTF boundries using the chain files  and 3) and add appropriate suffix to chromosome names (in genome and gtf/gff) and in geneid, transcript id in GTF/GFF files.

Using STAR ran the alignment on the diploid genome/gff-gtf. And obtained unique matches upto 42% which can be considered good given I had unique alignment to about 88% on haploid genome. And alignments to two loci on the diploid is about 48 %.

This is the parameter I used:


STAR --runThreadN 16 --runMode alignReads --genomeDir 1sTpassGenomeDir --readFilesIn chr2_R1.fastq chr2_R2.fastq --outFileNamePrefix FirstAlignment/2ms04h --outFilterMultimapNmax 6 --outSAMmapqUnique 60 --outSAMtype BAM SortedByCoordinate --outReadsUnmapped Fastx --outSAMattributes All --alignIntronMin 10 --quantMode TranscriptomeSAM GeneCounts --outSAMprimaryFlag AllBestScore --outFilterMismatchNmax 20

Any, suggestions on how and which parameters can be changed to promote more unique alignments. Let me know if I should should provide any output file.

Thanks,

[Alexander Dobin]

Hi Bishwa,

the 42% of unique alignments seems actually too high for me. I usually get only 10-20% when mapping to the diploid genome, as most of the reads do not overlap the variants, and hence map equally well to both haplotypes. Note, that if you add unique mappers and 2-loci-multimappers, you will get 42+48=90%, close to 88% uniquely mapped to the reference.

Cheers

Alex