Agenda
Presentation on optimization and comparison of PLAC-seq and HiChIP protocols
(Slides are available at https://docs.google.com/viewer?a=v&pid=sites&srcid=NGRudWNsZW9tZS5vcmd8NGQtbnVjbGVvbWUtd2lraXxneDozZWVjOGY4YTk0Y2RhN2M1)
Quality metrics for “C” libraries:
Trans ratio (reads between different chromosomes): higher means more random ligations;
Long range cis ratio: higher means more information from the “C” library
Enrichment level (PLAC-seq/HiChIP only): because immunoprecipitation/capture is not perfect
Discussion
Hu Jin from DCIC presented another idea of enrichment metric besides ratios of on-site enrichment: use an enrichment ratio independent of peak calling results, use a method similar to the Gini-index to measure the overall enrichment. There is a software named CHANCE ChIP-Seq that may help to calculate this Gini-index-like metric.
The details of the metric: calculate the accumulated distribution curve for both the ChIP and the input, the input should be like a diagonal line and IP results should be a curve below the diagonal. From there we can go in a few different ways: 1) area between the curves (similar to Gini Index), 2) maximum distance between the curves (also used in the CHANCE paper), 3) fit an linear regression model to find where the fit doesn’t work anymore. 1) and 2) are quite popular methods.
Bing Ren commented that the purpose of this call is trying to define the quality control metrics for PLAC-Seq and HiChIP protocols and would like to conduct a survey about which labs are currently using these technologies. Bing suggested Miao to share the slides and also the latest protocols so that other 4DN members would be able to evaluate on their behalf.
Hongbo Yang is using PLAC-Seq and Bing suggested him to measure the same quality control metrics to compare with Miao’s results.
Valentina Botti (from Neugaberu lab at Yale) is also using PLAC-Seq.
Edward Oh asked what is the approximate minimum sequencing depth required to calculate reliable target rate and fold enrichment value for PLAC-Seq.
Miao responded that she have used samples ranging from 6 million to 100 million reads and were getting similar results, so 5~6 million should be OK. Also it may depend on the number of peaks (Miao’s analysis involved several thousand of the peaks).
Feng Yu asked what’s the minimum amount of IPed DNA that she would start to make a library.
Miao replied that she could go as low as 5~10ng DNA. The amount depends on the complexity of the DNA but 10ng should be fine and she has no experience for a lower amount.
Feng followed that if the IP efficiency is only 0.5% for PLAC-Seq but the yield is higher than 10ng, would Miao still recommend to make a library.
Miao replied that low IP efficiency (IP yield) may actually be a good thing, because a high IP yield may indicates that the antibody is pulling down a lot of background information or the antibody is not as good. 10ng should be enough for a library.
Bing Ren would like to know the minimal fold of enrichment that members would feel comfortable for an PLAC-Seq/HiChIP experiment.
Hu Jin wondered that whether higher enrichment has a purpose because other “background” reads still contains Hi-C signal, which may also be valuable.
Bing replied that the reason of PLAC-Seq/HiChIP differs from regular Hi-C because they focus on interactions centered on certain regions (such as transcriptional factor binding site). So that may add value and save cost. And the purpose of the comment is trying to know the added value of PLAC-Seq and HiChIP. For example, when calling loops, PLAC-Seq/HiChIP may save a lot of reads than plain Hi-C.