search

Novartis / EQP-cluster

Unix-based RNA-seq quantification pipeline
Apache License 2.0
10 stars 3 forks source link

some values in assay are not integers #4

Open kubracelikbas opened 4 years ago

kubracelikbas commented 4 years ago

Hi,

When I try to analyze exon counts generated by EQP with DESeq2 package I get an error:

"Error in DESeqDataSet(se, design = design, ignoreRank) : some values in assay are not integers"

And when I go and check the results, indeed there are some decimal number. Why EQP produces counts that are decimal numbers and how I can fix this problem ?

Thanks.

sven0schuierer commented 4 years ago

Hi Kubra,

This is correct as EQP assigns fractional values to multi-mapping reads, for instance, a value of 1/3 to a read that maps to three locations in the genome. But this does not mean that the values produced by EQP are normalized in any way (e.g. by read depth to yield CPM). Since DESeq expects integer values, the solution is just to round the values of EQP to the next integer (this will lead to a negligible error of at most half a read for a gene).

Best regards,

Sven

From: kubracelikbas notifications@github.com Sent: Tuesday, December 24, 2019 7:51 AM To: Novartis/EQP-cluster EQP-cluster@noreply.github.com Cc: Subscribed subscribed@noreply.github.com Subject: [Novartis/EQP-cluster] some values in assay are not integers (#4)

Hi,

When I try to analyze exon counts generated by EQP with DESeq2 package I get an error:

"Error in DESeqDataSet(se, design = design, ignoreRank) : some values in assay are not integers"

And when I go and check the results, indeed there are some decimal number. Why EQP produces counts that are decimal numbers and how I can fix this problem ?

Thanks.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHubhttps://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_Novartis_EQP-2Dcluster_issues_4-3Femail-5Fsource-3Dnotifications-26email-5Ftoken-3DAC6VOKY7SETTDIOFX7XU3SDQ2GWLPA5CNFSM4J63J4P2YY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4ICOQFHQ&d=DwMCaQ&c=ZbgFmJjg4pdtrnL2HUJUDw&r=pN0-PFWjvy9WjMvRxmauk2R5qwVGB0pbr8n9_rKbibo&m=ez3GzCvYY9fHqmM7M4yJfT9oGn7mmgMZEsOwyrWPkvA&s=W_bpG72PF3nGZDvwAbssBVSTUXkmNGvSckgvoANRhrY&e=, or unsubscribehttps://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_notifications_unsubscribe-2Dauth_AC6VOK6A7QH6ZMHZCZVVZ5DQ2GWLPANCNFSM4J63J4PQ&d=DwMCaQ&c=ZbgFmJjg4pdtrnL2HUJUDw&r=pN0-PFWjvy9WjMvRxmauk2R5qwVGB0pbr8n9_rKbibo&m=ez3GzCvYY9fHqmM7M4yJfT9oGn7mmgMZEsOwyrWPkvA&s=2cBqAwAT9sPzN8xvsSq7_L4a7CYBfH_hcaxX_tBKrZM&e=.

Kelvin-Zeng commented 4 years ago

Hi, I encountered the same problem but it has been solved! DESeq2 should be run on the unnormalized count data. If you would like to use DESeq2 package, please make sure you initialize the object with a count matrix.