Star 2Pass alignment (2.5.0c)

[bassu]

Hi,

I am want  to do a 2Pass method alignment to human genome(Primary assembly fasta). I have multiple samples and  I'm confused(cant understand) with the help document/other sites on the steps involved.

This is what I understand

1) Create a Star index of the reference fasta file

2) Align each sample individually to the above index and obtain the SJ.out.tab

STAR --genomeDir Index/ --readFilesIn Sample1/R1.fastq Sample1/R2.fastq  --outFileNamePrefix Sample1

:
:

:

STAR --genomeDir Index/ --readFilesIn SampleN/R1.fastq SampleN/R2.fastq -outFileNamePrefix SampleN

3) Creating  new index with previous reference fasta and SJ.out.tab from all samples

STAR  --runMode genomeGenerate --genomeDir NewIndex/  --genomeFastaFiles Reference.fa --sjdbFileChrStartEnd  Sample1/SJ.out.tab  ... SampleN/SJ.out.tab

4) Remapping to the New Index

STAR --genomeDir NEW_Index/ --readFilesIn Sample1/R1.fastq Sample1/R2.fastq  --outFileNamePrefix Sample1

:
:

:

STAR --genomeDir NEW_Index/ --readFilesIn SampleN/R1.fastq SampleN/R2.fastq -outFileNamePrefix SampleN

Please do let me know if my above steps is correct?

[Alexander Dobin]

Hi @bassu,

your steps are correct. You may want to add annotations (GTF) files for both steps 1 and 3.

Also, you may want to filter the junctions in SJ.out.tab - for instance, you can remove chrM junctions and poorly supported non-canonical junctions.

Cheers

Alex

[bassu]

In GATK method they are mentioning not to provide a known annotation(https://www.broadinstitute.org/gatk/guide/article?id=3891).

Does this effect the results by any means? What would be the differences by giving a known GTF vs by not giving any?

[Alexander Dobin]

Hi @bassu,

the annotations are generally helpful, though may not be necessary for the SNP calling purposes, since they affect only mapping of low-abundance splices.

Cheers

Alex

[Yifang Tan]

Hi Alex:
I moved my question to the other threads, in case you do not come to this one anymore as it is marked completed.

Yifang