miRDeep2 output scores

Question

[Emilio Marmol]

Hello everyone, I'm using miRDeep2 to predict novel miRNAs in a Differential Expresion experiment from some muscle samples.

The thing is that when running miRDeep2, everything goes fine, but the problem comes when analyzing the results.

Does anyone know how to interpret the miRDeep2 scores?. The program generates a table with a header which tells you that the greater the miRDeep2 score is, the greater the probability is that this novel miRNA es a false positive, so by this, I understand that what is valuable to predict a good novel miRNA is that this miRNA has a low mirDeep2 score, near 0 or negative.

BUT, when going into the list of novel predicted miRNAs, which are designated by their coordinates, I find that most, if not all, of Differentially expressed predicted novel miRNAs, appear to have very high miRDeep2 scores, so according to what I mentioned above, those should have a very high probability of being false positives. BUT, as a variable within this list of novel predicted miRNAs, there's a field telling you the "estimated probability that the miRNA candidate is a true positive" which tells you that the higher the miRDeep2 is, the higher this probability of being a true positive is.

I find it conflictive, as firstly miRDeep2 score seems to be desirable if having a low value, but then it turns to the inverse interpretation, as it should have a high value.

Did someone analyze results from miRDeep2? I'm trying to select some desirable novel miRNAs to validate them via rt-qPCR but this is blowing my mind...

Here an example of what I get:

miRDeep2 score           estimated false positives
10                          13 +/- 2 (86 +/- 10%)
9                           14 +/- 2 (86 +/- 10%)
8                           16 +/- 2 (88 +/- 9%)
7                           16 +/- 2 (87 +/- 9%)
6                           16 +/- 2 (87 +/- 9%)
5                           19 +/- 2 (88 +/- 7%)
4                           19 +/- 2 (88 +/- 7%)
3                           23 +/- 2 (89 +/- 6%)
2                           39 +/- 2 (87 +/- 5%)
1                           33 +/- 4 (56 +/- 7%)
0                           0 +/- 0 (0 +/- 0%)
-1                          0 +/- 0 (0 +/- 0%)
-2                          0 +/- 0 (0 +/- 0%)
-3                          0 +/- 0 (0 +/- 0%)
-4                          0 +/- 0 (0 +/- 0%)
-5                          0 +/- 0 (0 +/- 0%)
-6                          0 +/- 0 (0 +/- 0%)
-7                          0 +/- 0 (0 +/- 0%)
-8                          0 +/- 0 (0 +/- 0%)
-9                          0 +/- 0 (0 +/- 0%)
-10                         0 +/- 0 (0 +/- 0%)

   id       miRDeep2_score    probability_that_the_miRNA_is_a_true_positive
9_5757           24516.9                     86 +/- 10%
12_566           459.7                        86 +/- 10%
17_2130          327.8                          86 +/- 10%
9_5807           313.3                        86 +/- 10%
X_6076          150.7                         86 +/- 10%
6_4694          150.3                         86 +/- 10%
14_1129         90.5                          86 +/- 10%
3_3571          22.6                          86 +/- 10%
1_2436          22                             86 +/- 10%
17_2140        15.5                        86 +/- 10%
X_6084          13.2                           86 +/- 10%
9_5766          12.8                          86 +/- 10%
2_3166          11.7                           86 +/- 10%
12_365           10                           86 +/- 10%
5_4071           9.5                              86 +/- 10%
12_306           8.1                              88 +/- 9%
2_3268           8                            88 +/- 9%
4_3853          5.4                          88 +/- 7%
4_3857          5.2                          88 +/- 7%
5_4248          1.9                          56 +/- 7%
17_1994         1.6                          56 +/- 7%
X_6116          1.6                         56 +/- 7%
X_6078          1.6                          56 +/- 7%
1_2408          1.5                          56 +/- 7%
X_6079          1.5                          56 +/- 7%
12_366          1.4                          56 +/- 7%
6_4829          1.4                          56 +/- 7%
2_3131          1.3                          56 +/- 7%
7_5135          1.3                          56 +/- 7%
5_4297          1.2                          56 +/- 7%
6_4818          1.2                          56 +/- 7%
6_4748          1.1                          56 +/- 7%
14_1223         1                            56 +/- 7%
7_5253          0.8                          0 +/- 0%
13_932          0.8                          0 +/- 0%
3_3413          0.7                          0 +/- 0%
3_3438          0.7                          0 +/- 0%

[biofalconch]

First off I would like to invite you to read the papers on both the original miRDeep and miRDeep2. I don't understand why you would think if a miRNA is higher differentially expressed it would be a false negative, and miRDeep does not care. miRDeep2 bases it search in these conditions:

1. Fold into Hairpin
2. Has mature, star miRNA and loop from hairpin in the smRNA-seq data
3. Have minimum 60% of the nucleotides in the candidate mature part base paired
4. Passing randfold threshold

After this miRDeep evaluates them and discard according to the set score threshold.

[henry-keen]

The other links don't work. Here is a more current link. I've read the paper, it only says that a higher score is better, there is no guidance, as far I can tell, about what score threshold should be used ( > 4, > 10, etc.)

https://currentprotocols.onlinelibrary.wiley.com/doi/10.1002/0471250953.bi1210s36

[evelinetavares]

Hello v82masae,

I have the same question on mirdeep2 scores. I just cannot figure out exactly what they mean. Have you got piece of the answer so far?

(sorry, i cannot help you. Maybe you got to an answer by now...)

Answer 1

[chupalav]

Same problem here. Stiil don`t know what an adequate score treshold is. Can anyone explain in more detail, please?