Drosophila differential expression.

Question

[vm.higareda]

I am working with a transcriptome of Drosophila melanogaster.

I am using the Hisat2_cufflinks_cuffmerge_cuffdiff methodology however I am having a lot of problems in each issue.

Right now I am confused about remove one replicate of four becouse it seem to be a outlier acording to the genes.read_group_tracking outfile.

If i remove this my results changue and now I have another outlier replicate.

I can understand this and others issues of cufflinks.

Is there someone doing this type of analisis ?

Answer 1

[Satyajeet Khare]

If your replicates are good, they will cluster together in dendogram if you plot in cummeRbund. Don't remove a sample just because "it seems to be an outlier". I omit specific samples in case of bad read quality or other issues such as overamplification or shoulder peaks on GC content graph on fastqc. If you are suspicious about a particular sample, you can go back and find out if there was any problem in its preparation.

[vm.higareda]

Hello

I do that and I got this image, dendogram, https://s18.postimg.org/5xn2g3lh5/file_page1.jpg

The labels " con_spiro " and " sin_spiro " are my treatments , but they are are not grouping in the same "clade

[Satyajeet Khare]

Two of your samples do look like outliers. You can omit them but if you can find out what happened, it will be great.