Uppercase vs lowercase letters in reference genome

Question

I am using a reference genome for mm10 mouse downloaded from NCBI, and would like to understand in greater detail the difference between lowercase and uppercase letters, which make up roughly equal parts of the genome. I understand that N is used for 'hard masking' (areas in the genome that could not be assembled) and lowercase letters for 'soft masking' in repeat regions.

1. What does this soft masking actually mean?
2. How confident can I be about the sequence in these regions?
3. What does a lowercase n represent?

Answer 1

What does this soft masking actually mean?

A lot of the sequence in genomes are repetitive. Human genome, for example, has (at least) two-third repetitive elements.[1].

These repetitive elements are soft-masked by converting the upper case letters to lower case. An important use-case of these soft-masked bases will be in homology searches: An atatatatatat will tend to appear both in human and mouse genomes but is likely non-homologous.

How confident can I be about the sequence in these regions?

As you can be about in non soft-masked based positions. Soft-masking is done after determining portions in the genome that are likely repetitive. There is no uncertainty whether a particular base is 'A' or 'G', just that it is part of a repeat and hence should be represented as an 'a'.

What does a lowercase n represent?

UCSC uses Tandom Repeat Finder and RepeatMasker for soft-masking potential repeats. NCBI most likely uses TANTAN. 'N's represents no sequence information is available for that base. It being replaced by 'n' is likely an artifact of the repeat-masking software where it soft-masks an 'N' by an 'n' to indicate that portion of the genome is likely a repeat too.

[1] http://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1002384

Answer 2

The use of lower/upper case letters and N/n letters in genomes sequences is not completely standardised and you should always check the specification of the resource you are using.

Lower case letters are most commonly used to represent “soft-masked sequences”, a convention popularised by RepeatMasker, where interspersed repeats (which covers transposons, retrotransposons and processed pseudogenes) and low complexity sequences are marked with lower case letters. Note that larger repeats, such as sizable tandem repeats, segmental duplications, and whole gene duplications are not generally masked.

However, there are other uses for lower/upper case letters, for example, Ensembl have used upper/lower case letters to represent exonic and intronic sequences respectively.

N and n nucleotides may represent “hard masked sequences”, where interspersed repeats and low complexity sequences are replaced by Ns. But N/ns may alternatively represent ambiguous nucleotides, indeed this is the IUPAC specification.

Also note occasionally (although fortunately rarely) X/x is used to represent ambiguous nucleotides or “hard-masked sequences” too.

Answer 3

1. Lower case nucleotides commonly denotes a soft masked sequences. How exactly the genome was masked you can find in FAQ of NCBI:

Are repetitive sequences in eukaryotic genomes masked?

Repetitive sequences in eukaryotic genome assembly sequence files, as identified by WindowMasker, have been masked to lower-case.

The location and identity of repeats found by RepeatMasker are also provided in a separate file. These spans could be used to mask the genomic sequences if desired. Be aware, however, that many less studied organisms do not have good repeat libraries available for RepeatMasker to use.

2. IMHO, low complexity regions are always more likely to be missassembled than high complexity sequences. However, this will be problem for non-model organisms. I would guess that the reliability of the softmasked regions of mouse genome will very high.

3. No idea, looks like an artefact.

An example of usage of the soft mask

Mapping of sequence to reference usually starts with perfect matches of seeds (substrings) of the mapped reads and the reference sequence. Soft masked (low complexity) regions are not used for matches of seeds, but they are used only for the extension of the alignment if there was a seed in a neighbouring region. This application of softmasking applied to problem of long read assembly is described on this blog.