giggle combo score 0

54 views
Skip to first unread message

angela...@gmail.com

unread,
Aug 5, 2019, 6:19:56 AM8/5/19
to giggle
Hi,

could you, please, someone help me? Do you have any idea why giggle results in combo score exactly 0 sometimes? We have downloaded several ChIPseq data for comparison, and in some case we have narrow and broad peak data from the same experiment, narrow results in high positive or negative combo score values, while the corresponding broadPeak file results in exactly 0. For me it seems to be an artifact, or bug???

Thanks for all your suggestions in advance,

Angela

Ryan Layer

unread,
Aug 5, 2019, 10:14:03 PM8/5/19
to angela...@gmail.com, giggle
Can you give some more info about the query, database, and command?
--
You received this message because you are subscribed to the Google Groups "giggle" group.
To unsubscribe from this group and stop receiving emails from it, send an email to giggle-discus...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/giggle-discuss/40252629-64f9-4076-a86c-951959ea97bb%40googlegroups.com.

Angela Bekesi

unread,
Aug 8, 2019, 10:30:41 AM8/8/19
to giggle
Hi Ryan,

I've sent you a reply in mail with details you asked for (for some reason it is not appearing here in the group page) you might not receive it...

Meanwhile, I've got an idea: I think my problem is similar to another issue arose here in the group conversation. My query files contain broad regions and can result in higher number of overlaps, than the number of regions in the query. Did I understand correctly that in such cases the p value might result in 0? Could it result in giggle combo score 0? If so, do you have any suggestion to solve the problem?

Thanks for your help,
Angela

On Tuesday, August 6, 2019 at 4:14:03 AM UTC+2, Ryan Layer wrote:
Can you give some more info about the query, database, and command?

On Aug 5, 2019, at 3:19 AM, angela...@gmail.com wrote:

Hi,

could you, please, someone help me? Do you have any idea why giggle results in combo score exactly 0 sometimes? We have downloaded several ChIPseq data for comparison, and in some case we have narrow and broad peak data from the same experiment, narrow results in high positive or negative combo score values, while the corresponding broadPeak file results in exactly 0. For me it seems to be an artifact, or bug???

Thanks for all your suggestions in advance,

Angela

--
You received this message because you are subscribed to the Google Groups "giggle" group.
To unsubscribe from this group and stop receiving emails from it, send an email to giggle-...@googlegroups.com.

Ryan Layer

unread,
Aug 8, 2019, 10:57:14 AM8/8/19
to Angela Bekesi, giggle
That could be the issue. Can you give the full line from giggle for one or the queries? I am a little concerned that there is a rounding issue.
To unsubscribe from this group and stop receiving emails from it, send an email to giggle-discus...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/giggle-discuss/28e3de9e-61ad-4f56-bfe9-78ab62deec83%40googlegroups.com.

Angela Bekesi

unread,
Aug 8, 2019, 11:00:10 AM8/8/19
to giggle
file_size    overlaps    odds_ratio    fishers_two_tail    fishers_left_tail    fishers_right_tail    combo_score
19838    49782    2.27E-10    1.17E-202    1.17E-202    1    0

Angela Bekesi

unread,
Aug 8, 2019, 11:06:38 AM8/8/19
to giggle
maybe the results of bedtools jaccard can provide additional information for you:
intersection    union-intersection    jaccard    n_intersections
270214637    837705883    0.322565    49629


On Thursday, August 8, 2019 at 4:57:14 PM UTC+2, Ryan Layer wrote:

Ryan Layer

unread,
Aug 8, 2019, 12:57:22 PM8/8/19
to Angela Bekesi, giggle
Ya, it’s an underflow issue. You have 20k intervals and 50k overlaps, which gives a fishers test of 1.1x10-202, which is pretty close to zero. I thought that we handled this case but we clearly don’t. Can you submit an issue?
To unsubscribe from this group and stop receiving emails from it, send an email to giggle-discus...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/giggle-discuss/350f9285-d8a1-4d34-8496-c352255d6076%40googlegroups.com.

Angela Bekesi

unread,
Aug 10, 2019, 10:43:54 AM8/10/19
to giggle
Hi, I could submit an issue for sure, if you help me what it means exactly, and how to submit... Is that something "official"?
Sorry, I'm pretty new in bioinformatics.
Reply all
Reply to author
Forward
0 new messages