STAR index file for GRCH37

Question

[bioinformatics.queries]

Hello everyone

I am trying to generate the index file for STAR alignment using hg19 genome. I used the following commad

STAR  --runThreadN 30    --runMode genomeGenerate  --genomeDir /data/shilpia2/STAR.index/ --genomeFastaFiles /data/shilpia2/STAR.index/GRCh37.primary_assembly.genome.fa --sjdbGTFfile /data/shilpia2/gff/gencode.v24.basic.annotation.gtf  --sjdbOverhang 100 --limitGenomeGenerateRAM 30000000000  --outFileNamePrefix /data/shilpia2/STAR.index/hg19

However, the program stops after a while without giving any error and without generating the index file. Could anyone suggest me what could be the reason or is there any problem in my command.

Thanks

[Barry Digby]

I would drop the --limitGenomeGenerateRAM and --outFileNamePrefix flags You could reduce --runThreadN to say, 8, (it might be a resource issue with your cluster). Also make sure that the --genomeDir exists. Let me know how you get on

[GenoMax]

How much memory do you have? You need at least 30G+ RAM for the index generation.

[bioinformatics.queries]

Thank you so much for your response. I used 30GM RAM to run my program and run it for 3 days but it still did not generate the file. Do you think i should run for longer time.

[GenoMax]

Did you have 30 cores available for the job? Did you get anything in log/error log?

Alex has pre-made hg19/GRCh37 indexes available at this link, if you can't make them.

[bioinformatics.queries]

I do have 30 cores available. The log file generated does not show any error. The running of STAR terminates after reading of the gtf file. I tried to use the index file from the link you provided. But it shows some error in the genome file.

[bioinformatics.queries]

This is what it appears in the log file.

 ..... processing annotations GTF
!!!!! WARNING: while processing sjdbGTFfile=/data/shilpia2/gff/gencode.v24.basic.annotation.gtf, line:
chr3    HAVANA  exon    198024658   198024788   .   +   .   gene_id "ENSG00000185621.11"; transcript_id "ENST00000482695.5"; gene_type "protein_coding"; gene_status "KNOWN"; gene_name "LMLN"; transcript_type "protein_coding"; transcript_status "KNOWN"; transcript_name "LMLN-002"; exon_number 15; exon_id "ENSE00003689636.1"; level 2; protein_id "ENSP00000418324.1"; tag "basic"; transcript_support_level "1"; tag "appris_alternative_2"; havana_gene "OTTHUMG00000155375.2"; havana_transcript "OTTHUMT00000339702.1";
 exon end = 198024788 is larger than the chromosome chr3 length = 198022430 , will skip this exon

[benformatics]

https://www.gencodegenes.org/human/release_24.html has a file named 'gencode.v24.basic.annotation.gtf'

There are all for hg38 not hg37/hg19.

The hg37/hg19 versions are here: https://www.gencodegenes.org/human/release_24lift37.html

[bioinformatics.queries]

From the link you provided should i download Comprehensive gene annotation file for GTF and Genome sequence, primary assembly (GRCh37) files ?

[benformatics]

It's up to you and depends on the goals of your study. I primarily use annotations from ENSEMBL and am thus not familiar with the basic vs comprehensive gene annotations. I think you should probably be fine undertaking standard differential gene expression analysis with the basic set but some features could be missing.

[bioinformatics.queries]

I just have another question. Which is better for alignment. I know people have been recommending to use STAR, but what if I use Bowtie. I was just trying to compare both the tools and see how much is the difference. I was looking for your suggestion. I have to do simple differential gene analysis. So it is ok if I use Bowtie?

[benformatics]

No, Bowtie is used for genomic alignments (i.e. DNA), for transcriptomic alignments (RNA) most would recommend a splice-aware aligner like STAR but you could also use TopHat2 (which uses bowtie under the hood).

[bioinformatics.queries]

Ok. Thank you so much for your response.

Answer 1

[GenoMax]

Are you mixing/matching sequences/annotations by any chance? They are all for the same build?

[bioinformatics.queries]

Hi

I did mix the annotation which caused the problem. I got the index file generated using the right GTF file. Thank you so much for your response.