Problem in count Matrix from Featurecount

Question

[daffodil]

Hi, I ran the following command, but when I compared the resulting count matrix to the BAM file in IGV, I noticed an issue. In some regions, despite the presence of several reads, the feature counts show zero.

I would greatly appreciate your feedback or suggestions on this matter.

# Run featureCounts for Genebody
featureCounts -p --countReadPairs -a "$GENEBODY_SAF" -F SAF -o “$OUTPUT_DIR/${SAMPLE_NAME}_Genebody_counts_new.txt" \
    "$BAM"

[Bastien Hervé]

A complete example focusing on few genes with IGV captures and extraction of the mentioned reads from the BAM file, would be helpful to solve your question.

Probably the reads you see in IGV are antisense or the other pair of the fragment is missing.

Answer 1

[Carlo Yague]

By default, featureCounts will not count reads that can not be assigned unambiguously: multimapped reads (reads with 0 mapping quality) or reads overlapping multiple features (multiple "gene bodies" in your case). This can cause the discrepancy between IGV and featureCounts. To disable this behaviour, you can use the -M and/or -O options in featureCounts (use with caution).