Repository for data validaton between QmRLFS finder and R-loop tracker tools

Each folder contains source data for the validation:

• gene_data folder contains raw sequences in fasta format
• experimental_data folder contains DRIPc sequenced R-loops in fasta format
• rlooptracker_data folder contains R-loop tracker analyses in bedgraph format

The script provides the following metrics for validation:

• Accuracy
• Sensitivity
• Specificity
• Precision
• Matthews Correlation Coefficient

Given dataset should provide following stats:

Validating R-loop forming signal detecton with experimental data

Plus strand

Gene	RLFS predicted by DRIPc sequencing	RLFS predicted by R-loop tracker
Immunoglobulin	-	+
MYC	+	+
RHOH	-	+
ACTB	+	+
FMR1	+	+
SNRPN	+	+
HK2	+	+
CIRH1A	+	+
APOE	+	+
FHIT	+	+
PPM1D	+	+
TP53	-	+
JTB	-	-
PBX1	+	+

Accuracy: 78.57 %
Specificity: 25.00 %
Sensitivity: 100.00 %
Precision: 76.92 %
Matthews Correlation Coefficient: 0.44

The detection algorithm produces more False positives compared to experimental methods which can be seen in Specificity, but overall performance and accuracy is adequate.

Minus strand

Gene	RLFS predicted by DRIPc sequencing	RLFS predicted by R-loop tracker
Immunoglobulin	-	-
MYC	-	+
RHOH	+	+
ACTB	+	+
FMR1	-	-
SNRPN	+	-
HK2	+	+
CIRH1A	+	+
APOE	+	+
FHIT	+	+
PPM1D	-	+
TP53	+	+
JTB	+	-
PBX1	-	+

Accuracy: 64.29 %
Specificity: 40.00 %
Sensitivity: 77.78 %
Precision: 70.00 %
Matthews Correlation Coefficient: 0.19

Results for negative strand are bit worse than for the positive strand and interestingly the algorithmic detection produces more False negatives.