I'm performing RNA-Seq data analysis for differentially expressed genes. as I'm new to this kind of work, so I performed these three gene count tools separately on the same bam file from RNA-Star alignment. I assumed that I might get a slite variation between these three tools' results. and I got some P-adj value variation for the same genes. so I want to know which one of these is better statiscally.
Different tools will, predictably, produce different adjusted p-values, but the total number of DEG should be fairly similar. For single-ended reads, featureCounts and htseq-count are nearly equivalent, but for pair-ended reads, featureCounts is more advanced. If you are unsure, compare the results of these tools (significant DEGs) and consider their intersection, even if the p-adjusted values are not the same but are lower than a significant threshold.
You can go for string tie for accurate reconstruction and better expression
If you are not limited to these three tools I would recommend you to look at Salmon (https://www.nature.com/articles/nmeth.4197 ) or Kalisto (https://www.nature.com/articles/nbt.3519). You can always scan the existing literature to see the comparisons, for example like this paper Article Comparative evaluation of full-length isoform quantification...
.
- Badges
- Science method