junction bed file from STAR

[frymor]

Hi everybody,

I am trying to create a junction file from the SJ.out.tab file of the STAR aligner.
I would like to see only those alignment in there, which have at least 5 reads covering the junction for canonical motives and 8 reads for non-canonical.

To get the reads I have used the following parameters:

--outFilterType BySJout
--outSJfilterCountTotalMin 10 8 8 8 
--outSJfilterReads default: All
--outSJfilterOverhangMin 30 12 12 12
--alignSJoverhangMin 8 
--alignSJDBoverhangMin 5
--outFilterMultimapNmax 20

I have than run STAR and modify the SJ.out.tab file in the following way:

awk {'if($4=="2") print "chr"$1"\t"$2-$9"\t"$3+$9"\tJUNC000"NR"\t"$7+$8"\t-\t"$2-$9"\t"$3+$9"\t255,0,0\t2\t"$9","$9"\t","0,"$3-$2+$9; else if($4=="1") print "chr"$1"\t"$2-$9"\t"$3+$9"\tJUNC000"NR"\t"$7+$8"\t+\t"$2-$9"\t"$3+$9"\t0,0,255\t2\t"$9","$9"\t","0,"$3-$2+$9'} $file >> $NEW

so that I get a bed file structure.

track name="IFM16h_1.bed" description="IFM16h_1.STAR.SJ.out.tab" visibility=2 useScore=1
chr2L	8082	8227	JUNC0001	26	+	8082	8227	0,0,255	2	35,35	 0,110
chr2L	8105	8240	JUNC0002	1	+	8105	8240	0,0,255	2	12,12	 0,123
chr2L	11330	11424	JUNC0003	2	-	11330	11424	255,0,0	2	15,15	 0,79

But even though I have tried everything to set a threshold for the "score" (the columns 7 and 8 of the original SJ.out.tab file, $7 and $8 in the awk command) greater than 1 I still get the in my junction file.

I would like to know if there is a way of making sure, that only splice junction with a coverage >=5 (or 8 ) will be in the SJ.out.tab file.

Does it means that the junctions with a coverage below 5 are annotated junctions? (If I understand the parameters correctly)

Thanks

Assa

[krespim]

Hi everybody,

I am trying to create a junction file from the SJ.out.tab file of the STAR aligner.
I would like to see only those alignment in there, which have at least 5 reads covering the junction for canonical motives and 8 reads for non-canonical.

To get the reads I have used the following parameters:

--outFilterType BySJout
--outSJfilterCountTotalMin 10 8 8 8 
--outSJfilterReads default: All
--outSJfilterOverhangMin 30 12 12 12
--alignSJoverhangMin 8 
--alignSJDBoverhangMin 5
--outFilterMultimapNmax 20

I have than run STAR and modify the SJ.out.tab file in the following way:

awk {'if($4=="2") print "chr"$1"\t"$2-$9"\t"$3+$9"\tJUNC000"NR"\t"$7+$8"\t-\t"$2-$9"\t"$3+$9"\t255,0,0\t2\t"$9","$9"\t","0,"$3-$2+$9; else if($4=="1") print "chr"$1"\t"$2-$9"\t"$3+$9"\tJUNC000"NR"\t"$7+$8"\t+\t"$2-$9"\t"$3+$9"\t0,0,255\t2\t"$9","$9"\t","0,"$3-$2+$9'} $file >> $NEW

so that I get a bed file structure.

track name="IFM16h_1.bed" description="IFM16h_1.STAR.SJ.out.tab" visibility=2 useScore=1
chr2L	8082	8227	JUNC0001	26	+	8082	8227	0,0,255	2	35,35	 0,110
chr2L	8105	8240	JUNC0002	1	+	8105	8240	0,0,255	2	12,12	 0,123
chr2L	11330	11424	JUNC0003	2	-	11330	11424	255,0,0	2	15,15	 0,79

But even though I have tried everything to set a threshold for the "score" (the columns 7 and 8 of the original SJ.out.tab file, $7 and $8 in the awk command) greater than 1 I still get the in my junction file.

I would like to know if there is a way of making sure, that only splice junction with a coverage >=5 (or 8 ) will be in the SJ.out.tab file.

Does it means that the junctions with a coverage below 5 are annotated junctions? (If I understand the parameters correctly)

Thanks

Assa

Hi Assa,

if I understood you properly the issue is that you can't filter out junctions with less than N reads (let's say N = 5) from the junction files. You can do it with awk before the junction file is even created:

awk '{if ($7 > 6) print $0}' $file

The condition {if ($5 > 6) print $0}' can be modified to include all sort of things, including the type site (canonical vs non-canonical)

awk '{if ($7 > 6 & $5 == 0) print $0}' $file

and variations of thereof. Also be careful that the SJ.out.tab is one-based and bed files are zero based. So in the output of your awk command the 3' of the 1st exon is actually the first 5' base of the intron. See this for a better explanation.