r3fang / snatac Goto Github PK

Hi there,

Really appreciate for your scATAC-seq pipeline! It really help my research a lot!
I have one question in your tutorial: I can't find mm10.fa file in your sample dataset after I unzip the mm10.tar.gz file. Could you please let me know where could I have this file? Thanks!

Simon

Hello,
I am trying to process the data. I am getting this error when using snATAC pre:

File "/home//anaconda3.4.3/lib/python3.6/gzip.py", line 260, in write
data = memoryview(data)
TypeError: memoryview: a bytes-like object is required, not 'str'

Hello !
The software ran successfully in the example dataset. but I noticed some of index sequence is different from the Supplementary Table 5 in paper 'Single_nucleus analysis of accessible chromatin in developing mouse forebrain reveals cell_type_specific transcriptional regulation'.
And with barcode sequence provided here, we can not completely correct barcode of dataset SRR6768121.
How could I generate index sequence according to the Supplementary Table 5 mentioned above ?
Do I need to reverse pcr index sequence or sth. else ?

Thank you and good luck !
Kylin

Hi there,

Really appreciate for your pipeline of snATAC-seq.
I follow your tutorial and execute the commands; however, some of the execution are quite slow (takes more than one day).
Could you please provide the output file in the step of

          bwa mem -t 1 mm10.fa \
          p56.rep1.R1.decomplex.fastq.gz \
          p56.rep1.R2.decomplex.fastq.gz \
          | samtools view -bS - > p56.rep1.bam

Also, is it possible to run this command in multiple cores?

Sincerely,
Simon

Hi, I am interested in the snATAC method, and I was trying to first reproduce your pipeline to make sure that I can do the same with my dataset.

I copied the tutorial exactly (except for downloading the mm10.fa, since I already have local BWA Index for mm10), and made sure that I am doing the same thing. However, when I filter out the barcodes with your pipeline (>1000 reads per cell, 3% consecutive promoter coverage, 20% reads in peak), I'm not getting 1464 barcodes--instead, 4600 of them--, and I am unable to produce any good clustering on the tSNE as it shows in the pipeline. I read in the paper that you got better clustering using the distal promoter regions, so I also tried creating a bed file containing regions +2kb upstream of the refSeq promoter.bed file as well, yet I still don't see much of a difference.

Would you be able to help me understand how you filtered out the barcode...? (+ tips on making good tSNE maps?)

Hi there,

I don't have the p56.rep1.txt file to proceed the intersectBed step after followed each of your previous codes. Do you have any suggestion? Thanks!

Sincerely,
Simon

Hi there,

I got the number 4598 from the command wc -l p56.rep1.xgi and
184519 from wc -l p56.rep1.ygi.
After I use the command ./bin/snATAC_jacard -i p56.rep1.mat -x 4598 -y 184519 -o p56.rep1.jacard
it shows: Segmentation fault

Do you have any suggestion?

Simon

Hi there,

I'm now working on the pipeline. I'm curious how do you generate the file "mm10_consecutive_promoters.bed"? If also need the file of hg19, where would be the best place to download them? Thanks!

Sincerely,
Simon

I couldn't see the source code of snATAC jacard, therefore I tried snATAC jacard with my test data.
However, I couldn't make out the results.

Input test data is as below (not resized peaks)

test.bed
chr1 1 1000 test_a
chr1 2001 3000 test_a
chr1 10001 20000 test_a
chr1 30001 40000 test_a
chr1 1 1000 test_b
chr1 10001 20000 test_b
chr1 10001 20000 test_c
chr1 30001 40000 test_c
test.xgi
test_a
test_b
test_c
test.ygi
chr1 1 1000
chr1 1001 2000
chr1 2001 3000
chr1 10001 20000
chr1 20001 30000
chr1 30001 40000

test.mat (this is the output of read.table(as.matrix("test.mat")) in R)
V1 V2 V3 V4 V5 V6
1 1 0 1 1 0 1
2 1 0 0 1 0 0
3 0 0 0 1 0 1

And the output of "snATAC jacard -i test.mat -x 3 -y 6 -o test.jaccard" was below
0.00 600.00 600.00
600.00 0.00 500.00
600.00 500.00 0.00

This result seems that the function "snATAC jaccard" calculate the Jaccard distance of the vector assigned to each cell (e.g. (1,0,1,1,0,1) for test_a cell) and we have to resize the width of each peak in .ygi.
However, if so, the Jaccard distance between test_a and test_b would be 0.5(2/4) and the value of 1000x jaccard would be 500. Why is the result is slightly different from the expected result?

(I think "1" is added both denominator and numerator. (2/4→3/5→0.6, 1/3→2/4→0.5) )

Hi, First - I wanted to say thanks and well done on establishing this protocol. This is very impressive and interesting! We are very interested in scATAC-seq of tissue samples and would like to look at your dataset. However, it seems that not all the files in the http://renlab.sdsc.edu/r3fang/snATAC/ folder are open for public download. For about half of them (e11.5_R2, e12.5_R2, e14.5_R2, e15.5_R2, e16.5_R2, p0_R2, p56_R1, p56_R2), the following error arises when trying to Download

-sh-4.1$ wget http://renlab.sdsc.edu/r3fang/snATAC/p56_R1.decomplex.fastq.gz
--2018-03-07 11:23:55--  http://renlab.sdsc.edu/r3fang/snATAC/p56_R1.decomplex.fastq.gz
Resolving renlab.sdsc.edu... 198.202.90.216
Connecting to renlab.sdsc.edu|198.202.90.216|:80... connected.
HTTP request sent, awaiting response... 403 Forbidden
2018-03-07 11:23:55 ERROR 403: Forbidden.

We wanted to look at the full dataset, would you be able to provide download access for all the decomplexed fastq files?

Many thanks, Julia