cuffdiff questions

[bvk]

@bvk: The error you are getting is likely not from cuffdiff. It appears to be coming from your python script. I recommend that you create a new thread and indicate that you need help with a python error in the tile.

ok. Thankyou

[GenoMax]

@bvk: The error you are getting is likely not from cuffdiff. It appears to be coming from your python script. I recommend that you create a new thread and indicate that you need help with a python error in the tile.

[bvk]

I think I can help with your first question.

Cuffdiff can find the differentially expressed genes between groups or across replicates depending on how you input your data.

For example:
Group1 has replicates A, B, C, D, and E
Group 2 has replicates 1, 2, 3, 4, 5

#all vs all comparison
$ cuffdiff transcripts.gtf A B C D E 1 2 3 4 5

#group 1 vs group 2 comparison
$ cuffdiff transcripts.gtf A,B,C,D,E 1,2,3,4,5

In the group vs group comparison the biological replicates are linked together by commas.

-Pete

I used this in command line:

cuffdiff -o diff_out4 -b ../genome/ce10.fa -p 2 -L larval,early -u merged_asm/merged.gtf ../tophat/em/SRR493359_60_61_thout/accepted_hits.bam ../tophat/em/SRR493363_64_65_thout/accepted_hits.bam

here SRR493359_60_61 is one group and SRR493363_64_65 is the other group. This worked for me.

Now I want to run this from a python script. So, I gave a call in this way.

do.call([cfg.tool_cmd("cuffdiff"), "-p", str(cfg.project["analysis"]["threads"]), "-b", str(cfg.project["genome"]["fasta"]), "-u", cfg.project["samples"][0]["files"]["merging_gtf"], "-L", str(cfg.project["phenotype"][0]), str(cfg.project["phenotype"][1]), "-o", output_folder] + [cfg.project["samples"][0]["files"]["bam"] cfg.project["samples"][1]["files"]["bam"]], cfg.project["analysis"]["log_file"])

here
str(cfg.project["phenotype"][0]) is larval
str(cfg.project["phenotype"][1]) is early

I get an error: invalid syntax. Can you please help in this. Thanks in Advance

[edge]

Hi pinki999,

Do you fix the problem already?
I'm facing the same error message as well
need your advice....
how to correct (--frag-len-mean and --frag-len-std-dev)?

[pinki999]

Hi,

I downloaded the new version of cufflinks and again tried to use cuffdiff. I just used the -N option for upper quartile normalization and my output files are again empty. I got the following warning messages

[09:52:55] Modeling fragment count overdispersion.
Warning: fragment count variances between replicates are all zero, reverting to Poisson model
Warning: Using default Gaussian distribution due to insufficient paired-end reads in open ranges. It is recommended that correct paramaters (--frag-len-mean and --frag-len-std-dev) be provided.
> Map Properties:
> Upper Quartile: 93.00
> Read Type: 50bp single-end
> Fragment Length Distribution: Truncated Gaussian (default)
> Default Mean: 200
> Default Std Dev: 80


And this was repeated quite a number of times with different upper quartile values.

Note: I am using the .gtf file downloaded from ftp://ftp.ensembl.org/pub/release-54/gtf/homo_sapiens/ (for the hg18 build)

Thank you

[DZhang]

Pinki, check all output files of cuffdiff. Did you go through cufflinks, cuffcompare, and cuffdiff? Please check all the outout (cufflinks - gtf), cuffcompare (gtf), and cuffdiff (_diff and _fpkm.tracking).

Douglas

https://www.contigexpress.com

[peromhc]

Pinki, what does the screen output tell you? Cufflinks output is usually pretty informative.

[pinki999]

Hi Peter,

Thank you for your reply. I used cuffdiff for group1 vs group2 analysis, but all my output files are empty

[pjorth]

I think I can help with your first question.

Cuffdiff can find the differentially expressed genes between groups or across replicates depending on how you input your data.

For example:
Group1 has replicates A, B, C, D, and E
Group 2 has replicates 1, 2, 3, 4, 5

#all vs all comparison
$ cuffdiff transcripts.gtf A B C D E 1 2 3 4 5

#group 1 vs group 2 comparison
$ cuffdiff transcripts.gtf A,B,C,D,E 1,2,3,4,5

In the group vs group comparison the biological replicates are linked together by commas.

-Pete

[pinki999]

Hi,

I am working on RNA-seq data (Solid) which contains 2 sample groups and each group contains 5 biological replicates. I am interested in finding the differentially expressed genes between these 2 groups. I want to use cuffdiff for this purpose. However, I have certain questions:

1) Does cuffdiff find differentially expressed genes between the groups or across the replicates?

2) Data can be normalized using the option -N (upper quartile normalization). I believe it carries out within sample normalization but what about sample to sample normalization? Does cuffdiff carries out sample to sample normalization before comparing the groups?

Can anyone kindly answer these questions?


Thank you