While checking for deferentially expressed miRNAs after deep sequencing, i got some strange values like 1.79769e+308 or 1.79769e+308, which i can't understood. The rest of the values are +1, +2, +3 or -1, -2, -3, and are easy to understand. Can anybody tell me what do other values mean, or how to adjust them with the rest of the values?
I am just guessing here, but that looks like the maximum possible double value.
This usually happens if some calculations is programmed wrong or sloppy or may be a choice of the programmer to indicate that there is something unexpected going on with the data.
If you are using a third party program, I suggest you design a synthetic dataset and first ensure that the program actually does what you expect it to do.
@thank you Jens Allmer; infact i did deep sequencing through a company and they provided this data
I agree with Jens. It may happen if the denominator is zero, so the fold-difference goes to infinity, what is seemingly substituted by the maximum double precision floating point value (possibly to avoid errors caused by "NaN" or "Inf" values). It might be interpreted as an "switch-on" effect (miRNA sequence is absent under the reference condition, and present to some extent under the experimental condition). However, it is a good idea to check this with Jens suggestion to analyze a synthetic dataset.
Hi Hamid,
this usually happens if the RPKM for one treatment is zero. You could replace all zeros with e.g. 0.001 to get more 'user-friendly' ratios, but I would keep this in mind and use a sensible cut-off in downstream analysis or visualisations.
@Jochen; yeah you are right, it's because of the absence of microRNA in reference condition hence the denominator is zero.
Now i want to know, as i am constructing a heat-map for deferentially expressed microRNAs and i have some values like this e.g. mir-124: Treatment 1: (log2fold change =1.11852), T2:(-1.82522), T3:(-1.79769e+308) T4:(1.79769e+308) T5: (0.937921) T6: (1.79769e+308)
so i don't know how to adjust these values or is it the right option to exclude it due to absence of microRNA reference value ?
No, don't exclude them. You can define some cut-off value for the color in the heatmaps (actually the same as "adjusting" the values, as you said). Then, for instance, a "bright red color" would indicate a "very strong induction or a switch-on event", something like this. It is absolutely sufficient to help recognizing patterns in the regulation - what is the intention of heatmaps.
Thank you, can you suggest some cut-off value in the case of above mentioned example?
plus/minus 2, since -1.8 was the largest "reasonable" effect you reported (T2).
depends on the rest of your miRNAs as well, you probably want to show a range that is biologically sensible (and stat. significant)
- Badges
- Science topic
- Similar topics
- Biological Science
- Biochemistry
- Biomolecules
- Nucleic Acids
- RNA
- Small RNA
- microRNA