Linear DNA amplification (LinDA). Look at the following article in Nature methods.

http://www.nature.com/nmeth/journal/v8/n7/full/nmeth.1626.html

It is easy to follow and the ChIP-seq results were nice and as comparable to what was done from millions of cells.

I have used WGA and LMPCR while performing ChIP-on-chip in the past but also managed to get a lot of background noise for LMPCR samples hybridised to the tiling arrays, never hybridized the WGA samples though.

Dear Daria,

Please go through the below given link. This may help you.

http://www.ncbi.nlm.nih.gov/pubmed/?term=23439889

or Please contact Dr. Raja CEO or Dr. Sudha, COO of Genotypic Technology they can able to provide you better suggestions. mail id. [email protected], [email protected]

This paper may be helpful.

In vivo epigenomic profiling of germ cells reveals germ cell molecular signatures

Since the library prep include an amplification step I wouldn't bother with amplifying the sample before, worse case you can add a few cycles to the library amplification. You should defiantly avoid any process that add adapters since that mean that all your reads would start with the same sequence which cause the Illumina sequencer to throw them away, we had a lot of truble even with a short lead when we had our barcodes on the 5' end before Illumina switched to the TruSeq system.

I have just recently analyzed ChIPseq data that was produced after LinDA amplification. Although the method sounds nice and easy, one has to make sure that the adapters used for LinDA are reliably cut and removed from the sample (especially if the read length is < 50bp). I have seen a lot of adapter contaminations and poly-A tails in that data.

Clean reads from Illumina sequencing: 24.932.016

Clean reads after filtering LinDA adapters: 12.430.636

However, the method might work better in your hands so it could be a good way.

This doesn't directly address your question, but if you are doing ChIP from low cell numbers, native ChIP may be useful:

http://www.biomedcentral.com/1471-2164/13/645

http://www.epigeneticsandchromatin.com/content/6/S1/P9

From what I can tell, these protocols don't require post-ChIP amplification.

Many thanks for all contributions! As we are not going to prepare the Illumina library ourselves, the only option is to go for amplification (the sequencing company does not agree to start from total 2 ng of material).

Dear Dietmar Rieder and Biswajyoti Sahu! as you seem to have direct hands-on experience with LinDA; I have additional questions to you, after reading the protocoll. In the final step, the authors 1) use 10U BpmI and HaeIII (NEB) to get rid of the primers, and than 2) directly go for Illumina sequencing.

1)I don't understand, why in the final step the authors use HaeIII, which recognizes 5'GGCC3'!? There is no such a sequence within the primer: TAATACGACTCACTATAGGGAGACTGGAG(A16) . Have you used HaeIII?

2) To my mind, going directly to Illumina sequencing (without purification from LinDA adapters, let us say, by PCR purification kit) - would result in adapter contamination, and Dietmar notes exactly this point. Have you cleaned up the last reaction before sequencing?

Hi Daria,

There is an explanation given in the following Nature Protocols from the same authors:

http://www.nature.com/nprot/journal/v7/n2/full/nprot.2011.447.html

Hae III digestion is optional, however we haven't used it and have only done the BpmI digestion and then purified using PCR purification kit. Of course in that case, we have to get read of the extra oligo (A) sequences. The authors have also provided a small script to do this.

One more thing we routinely do is perform a Bioanalyzer analysis before sequencing, in that case we can detect the adapter and can re-purfiy the sample if needed.

Biswajyoti Sahu is completely right. I would suggest to check with the BioAnalyzer and use Ampure or SPRIselect to do a final clean up.

hi all,

nice to see an animated discussion on LinDA, perhaps I can be some help in this matter, I am one of the authors.

Bpm1 digestion is to remove the T7 adapters from the ends as Illumina method sequences only the ends. We used HaeIII in the past to improve our coverage, however, with the advent of Hiseq2000, we have huge number of reads, and TrimLinDA, a script to cleanup the sequences, we dont need to use Hae III anymore.

Another improvement I can suggest to improve your yields is to use Ampure beads (SPRI, Beckman Coulter) to purify your dsDNA instead of column, you not only improve the yield but you get rid of the adapters better.

Dietmar, regarding your comment about the poly A tails in the reads, we have improved the script TrimLinDA thereby increasing the total reads recovered. it is on our lab website

http://archive.igbmc.fr/recherche/Prog_FGC/Eq_HGron/bioinfotools.html

Our bioinformatician has developed another novel approach to address this issue by targeting the alignment with Bowtie, he shall be writing a detailed answer soon

Pattabhiraman, thanks for the information. I initially used the TrimLinDA script and even improved it a bit an my own (e.g. match also the reverse complements of the adapters or adapter residuals, do not per default remove the first A). But then end I ended up using flexbar [1] and a tailored fasta file containing all possible contamination for adapter cleaning. When I checked the quality of the reads with fastqc this gave pretty good results.

I'll test for sure the new TrimLinDA script as well.

[1] http://sourceforge.net/projects/flexbar/

Hi all, I am from the same lab as Pattabhiraman. I have analyzed few LinDA amplified ChIP-Seq profiles. From my experience, instead of trimming the data in prior for adapters if we go for local alignment in bowtie2 yields very good results... this option is not available in bowtie...

In this local alignment mode (lik BLAST), the aligner itself will look for best longest match.. if there are no matches as a whole read, then the tool will try trimming from one or both the sides till it finds a best longest match... in this way we can let the tool to decide to find the best alignment by trimming unwanted adapters/mismatches towards the ends... Filtering uniquely mapped reads will take care of falsely aligned short reads or if needed one can filter short alignments from SAM file itself..

Recently, with this approach I have got more than 70% of reads uniquely mapped to the genome from a LinDA amplified data...

PS: This is not only for LinDA amplified data, it can be applied for any data where you expect some adapters at the ends...

Mohamed, this sound reasonable to me! Good suggestion. This will work well with long reads and is less of an effort. The adapters are still problematic if you have short reads

I agree.. But the percentage of reads containing whole adapter is very less, I guess. These reads are anyway not going to be helpful with any other trimming process too. That is a tradeoff which has to be chosen based on percentage of clonal reads from PCR amplified data and percentage of unusable reads from LinDA amplified data...

And like you said removing loose adapters using Ampure beads could be very helpful too...

Dear all,

Since you have such a lively discussion on the LinDA protocol here I decided to barge in with some more specific questions, hoping that it will also be helpful to the people who are considering or just starting to use this technique.

I am trying to optimize the protocol for several months now, but in my hands it does not work so well (I get only 40x amplification rate on average).

Working with PCR fragment I realized 2 things:

1. The T-tailing is not very effective. One can improve the efficiency of the reaction by increasing the Tmix concentration. Theoretically, since the proportion of ddCTP remains the same the length of the T-tail should not increase, Of course to visualize the fragment I had to start with large amount of DNA (30ng) where the Tmix might have been the limiting factor. Do you think this would hold true also for smaller DNA amounts?

2. The 2nd strand synthesis is not very efficient. Looking at the size distribution after the second strand synthesis I see that more than 60% of my DNA remains single stranded (i.e. contains only one primer). How do I improve this step? Do you think repeating the PCR cycle one more time would be wise?

I would be very grateful to all the people with hands on experience with this method for shearing their knowledge. Any other tips/suggestions on how to improve my amplification rates are very welcome.

hi alisa,

1. i would not recommend increasing the Tmix in the reaction. as you point out correctly the effect you observe is due to larger starting amts. if you were to increase the Tmix in normal LinDA you will have huge T tails and the seq would be very difficult.

2. i dont think i understood how you found out 60% of the dna is single stranded. Which primer did you use to do the PCR? could you please clarify.

3. you are quite right that the 2nd strand synthesis is inefficient but reverse transcription step also has to shoulder some of the blame. it is a mixture of 2 that gives you poor yields. but this is the best one can do, i have tried many different protocols but none are better.

Regarding what can be done

1. switch to SPRI purifications, the yield and the quality is much much better. it is very simple to use and very effective, so replace all column steps with it, RNA purif, 2nd strand purif and the post Bpm purif. (please see above posts)

2. be careful with your T mix and primers, not too many freeze thaw cycles and dedicated stuff for linda (it makes a big diff, believe me)

3. which CoCl2 are you using? DO NOT USE the 2.5 mM COCL2 provided with terminal transferase by NEB, please make a 5mM solution from powder or someplace else and use 1 µl in your reaction. this makes a huge difference, many people have noticed this effect.

hope it helps