Doubt related terms used in HOMER for ChipSeq Analysis

Question

[pinky_pinkpinky]

Dear All,

Can any one working with HOMER help me in understand few frequently used terms in HOMER.

I read the HOMER manual. I have doubts as to how to define these points.

Normalized tag count, Total tags, Control tags (normalized to IP), Fold change Vs Control, Fold change Vs Local, Clonal fold change

Thank you

[pinky_pinkpinky]

Thanks so much for your answers. Its true the manual has discussed well but I had a very fundamental doubt .

I believe tags are the reads while I was confused with the protein tagging in chip-seq.

Secondly as per your answer,

Fold change vs control is for the putative peaks where in, the fold change of normalized tag counts vs control tags is scored above threshold. So what is the threshold over here.

Is it similar to -F option which we specify . In such case I have used default 4.0

-F <#> (fold enrichment over input tag count, default: 4.0).

Kindly confirm it.

Thanks again for your kind reply.

[Manvendra Singh]

yes, by default the fold change for assigning peak is 4 unless you specify from your side in -F option

[Manvendra Singh]

No NO No, please do not get confused in the case of tag, this is not that tag by which we label proteins.

This is sequencing reads :)

Answer 1

[Manvendra Singh]

If I understood correctly then,

Normalized tag count is number of statistically significant tags in the defined peaks which normalized with number of total tags mapped (By default it normalizes with 10 million reads, unless you define some number in the option)

Total tags are the total number of tags mapped to genome (not to be confused with total number of tags contributing peaks)

control tags are from input (self explanatory)

Fold change vs control is for the putative peaks where in, the fold change of normalized tag counts vs control tags is scored above threshold

Fold change vs Local is where putative peaks are narrowed down after filtering the peaks on the basis of width of region as much you specify in the option -L <>

Clonal fold change is another filtering step, where it discards those peaks which are false positive over repetitive elements, It take observed unique regions vs expected unique regions covered by tags, and upon specifying the value in -C <> it discards the peaks above that value (fold change)

btw its well written in the manual.

HTH

[pinky_pinkpinky]

Thank you once again.

Can you please clarify on the term "statistically significant tags" which you mentioned in Normalized tag count is number of statistically significant tags in the defined peaks which normalized with number of total tags mapped.

How to consider the tags to be statistically significant?

Thanks

[Manvendra Singh]

It means that those peaks which are found above the threshold of "FDR", "p-value of Poisson's distribution", "peak size" etc.

whatever you specify or go for default options.

Its all back-ground corrected peaks which are detected by Homer's algorithm.

[pinky_pinkpinky]

Hi Manvendra,

I have one more doubt. I have Chip-seq results obtained through HOMER that has the chromosome num, start and end positions and normalized tag count, Focus ratio, Total tags, control tags etc. Can I use the normalized tag count value as the read count for that region (chr start and chr-end). I need to plot the enriched peaks through R. Is this a correct approach?

Thank You

[Manvendra Singh]

Yes, You can consider them as read counts

[pinky_pinkpinky]

Dear Manvendra,

I saw in many articles referring the bam files to get the read count and then generate the peak images with respect to it. In such case what is the use of normalized tag count information. Will it be wrong if I use the normalized tag count values for generating the peaks. I am slightly confused. I need a strong confirmation for the same.

Thanks,
Pinky