Error importing .hic file: File doesn't have the given chr_chr map 1_1.

[Charles-Alexandre-Roy]

Hello,

I am attempting to perform an analysis with OHCA but am running into an issue early on that I can't seem to debug.

First, I created a .pairs file following the pairs format specifications, e.g.:

...
#columns:	readID	chr1	pos1	chr2	pos2	strand1	strand2
.	chr1	103068099	Nadia_NL4-3	5486	.	.
.	chr1	107159350	Nadia_NL4-3	8403	.	.
...

Then I used Juicer tools to create a .hic contact matrix from the .pairs file, as described here.

Then, in an RStudio session running the OHCA tools, I ran the following commands (as described here), to try to generate a HiCExperiment object:

library(HiCExperiment)
library(HiContacts)
hic_fp <- HicFile("/path/to/contact_map.hic")

Running the following commands appear to work as expected:

availableResolutions(hic_fp)
resolutions(10): 1000 5000 ... 1e+06 2500000

availableChromosomes(hic_fp)
Seqinfo object with 712 sequences from an unspecified genome:
  seqnames         seqlengths isCircular genome
  chr1              248956422       <NA>   <NA>
  chr2              242193529       <NA>   <NA>
  chr3              198295559       <NA>   <NA>
  chr4              190214555       <NA>   <NA>
  chr5              181538259       <NA>   <NA>
  ...                     ...        ...    ...
  chrUn_KI270385v1        990       <NA>   <NA>
  chrUn_KI270423v1        981       <NA>   <NA>
  chrUn_KI270392v1        971       <NA>   <NA>
  chrUn_KI270394v1        970       <NA>   <NA>
  Nadia_NL4-3            9181       <NA>   <NA>

Then, if I run the generic import command, the session seems to crash and R becomes unresponsive:

hic <- import(hic_fp)

Note that I am running these commands on a server with 16 CPUs and 128 GB of memory, so I doubt that my machine is the issue.

If I run a command such as the following to try to limit the import, I get the error message following the command:

hic <- import(hic_fp, focus = "chr1", resolution = 2500000)
Error: File doesn't have the given chr_chr map 1_1.

Any idea what could be going wrong here?

[js2264]

Hi @Charles-Alexandre-Roy , thanks for opening this issue. Two comments:

• First, the high number of chromosomes (+ contigs) in your reference may actually be a bottleneck for importing the genome-wide contact map from a .hic file with straw. strawr (the underlying R library developed by the lab who came up with the .hic format) cannot extract contacts over more than a pair of 2 chromosomes, and 712 chromosomes/contigs mean 253116 combinations of any pair of 2 chromosomes/contigs... Do you really need the extra unnanotated contigs? If no, I would recommend removing them from your pairs file, then recomputing a .hic file. If you do need to keep them, then I would recommend parsing your pairs into a .cool file (eg using the cooler cload pairs ... CLI tool).
• I suspect the error you mention might come from the chromosome nomenclature used here. Unannotated contigs have "", and you also added an extra chromosome reference "Nadia_NL4-3", with "" and "-" characters. Perhaps try renaming "Nadia_NL4-3" to only have alphanumeric characters and filter out chromosomes with non-alphanumeric characters. I suspect juicer internally handles some chromosomes differently if some unexpected characters (eg "_" or "-") are found in some but not all chromosome names.

[Charles-Alexandre-Roy]

Hi Jacques, thanks so much for the help, I really appreciate it!

You were right, after removing all alt and unlocalized contigs from GRCh38, the import function was able to finish without crashing the session.

This helped reveal the source of the cryptic error messages like "Error: File doesn't have the given chr_chr map 1_1.". For some more context, the goal of this analysis is to examine the contacts between human chromosomes and a retrovirus genome that has become integrated into the human genome in single-cell data. All human-to-human contacts have been already filtered out of the dataset and do not appear in the .hic file I am importing. When I imported "everything" after removing the extra contigs, I noticed that only the virus-to-virus contacts were actually imported. When I run something like: hic <- import(hic_fp, focus = "chr1", resolution = 2500000), I still get the error, which seems to be stating (in a quite obscure way) that there are no chr1-to-chr1 interactions in the .hic file.

In the OHCA documentation, you mention that inter-chromosomal interactions can be defined. Is there a way of importing inter-chromosomal interactions in a .hic file directly?

[Charles-Alexandre-Roy]

@js2264 when you have a second, please let me know if you have any suggestions for resolving the above.