It is common misunderstanding of genetics to think that we are 50% like our mother and 50% like our father. It is true that we inherited half of our DNA from each parent, but that doesn't mean that the end product is 50-50.
Firstly, most of the human DNA is identical between two unrelated individuals. This is estimated to 99.5% for individuals from different ethnic groups, but the similarity could be as high as 99.9% for two persons whose ancestors all come from the same small region, especially in rural areas where the genetic pool is more limited. In relatively isolated provinces almost anybody within a few villages may very well be cousins within a 5 to 10 generations.
In the 50% of DNA we inherit from each parent, over 49% will be identical to the half inherited form the other parent. This is at least true for women, who have two X chromosomes and therefore have symmetrical pairs of chromosomes. Men receive a shorter Y chromosome from their father. The Y chromosome is about 2.5 times smaller than the X chromosome. So men inherit more DNA from their mothers. This may seem counter-intuitive, but that's the way it is.
Genetic similarity between siblings
However, a boy/man is likely to be genetically closer to any of his brother than his sisters or mother. This is because the Y-chromosome is identical between brothers, or from father to son.
Brothers won't have the same X chromosome though. People do not inherit a full X chromosome from their mother, but some segments of each of the mother's two X chromosome. Theoretically, two siblings could get the exact opposite (or same) combination of X segments, although the chances are very low.
Sisters are even closer to each others because they have one full identical X chromosome (inherited from their father), which is bigger than a Y chromosome. Their second X chromosome is also an random admixture.
The Y chromosome continuity
The only chromosome that survives the test of time is Y. All the other 22 chromosomes + X are recombined at each generation, so that you may not have any identical segment left when compared to a great-great-grand-parent. The Y chromosome remains virtually unchanged. A few minor mutations do occur at each generation, but on a sequence of 60 millions bases, it takes hundreds of generations before the change becomes significant.
Amazingly, our ancestors have realised long before the genetic era that the only long-term genetic continuity was passed from father to son. Surnames are also passed from father to son, and lost through the mother in most cultures, as if to reflect this Y-DNA continuity. It could have been a coincidence, but the fact is that family names are supposed to match Y-DNA in most non-tribal societies on Earth. One notable exception are cultures in which the surname change every generation based on the father's given name (e.g. in Iceland or Arabic countries).
More like mum or dad ?
The 0.5% to 0.1% of DNA that differs between two human beings are called alleles or SNP's (single nucleotide polymorphism). This is really the only thing that matters when comparing yourself to someone else.
The autosomal DNA (the 23 pairs of chromosomes minus X and Y) always works in pair. Wherever you look at the DNA sequence there will be two "letters" (each representing an amino acid). There only 4 letters in the DNA language : A, C, G and T. When the letter is the same on both chromosome at the same location (e.g. AA), it is said to be homozygous. When the letters differ (e.g. AG), it is heterozygous. This is important to understand how a child can be much more similar to one parent than to the other.
Imagine that you are looking at a DNA sequence on one gene known to have polymorphisms (i.e. variations from one person to another) in that 0.5% of DNA that is not identical in all human beings. A child's mother might have a sequence like this :
AA, CC, GG, AA, CC, CC, CC, TT (perfectly homozygous)
The father, on the other hand might have the exact opposite version of the gene, but also homozygous :
TT, GG, CC, TT, GG, GG, GG, AA
So any child the couple might have will always be heterozygous :
AT, CG, CG, AT, CG, CG, CG, AT
In other word, the child is just 50% like his father and just 50% like his mother, for that very short stretch of DNA.
But what if the mother is homozygous (same sequence as above) and the father is heterozygous ? In that case the child will either be homozygous like the mother, or heterozygous like the father. The child will get an A from the mother's AA, but either an A or a T from the father's AT. He/she will be AA like the mother or AT like the father. For that particular sequence, the child just cannot be 50% like each parent, but will be inevitably be 100% like the mother OR to the father, and 50% like the other parent.
Now what if both parents are heterozygous ? The child has 50% of chances of also being heterozygous (AT), but also 50% of chances of being homozygous (AA or TT). In that case the child is neither identical to the father, nor the mother, but half like each (but the same half).
To know if you are genetically closer to your father or mother (or anybody else), all the SNP's should be counted. For each base pair, you can either be 100% identical, 50% identical or 0% identical from someone else.
Let's take another example. Imagine that there are only 100 SNP's in the genome. When comparing your DNA to your parents, you get this :
- 60 SNP's that are 100% identical, 30 SNP's that are 50% identical, and 10 SNP's that are 0% identical with your father.
- 40 SNP's that are 100% identical, 40 SNP's that are 50% identical, and 20 SNP's that are 0% identical with your mother.
You get 2 points for the 100%, 1 point for the 50% and, 0 point for 0%.
120 + 30 = 150 points of similarity between you and your father
80 + 40 = 120 points of similarity between you and your mother
In this example, you are genetically closer to your father, despite inheriting only half of his DNA. This is possible because there are only three versions for any base pair in the genome (e.g. AA, AT or TT).
Companies like 23andMe already allow to compare yourself to any friend or family member who also takes the test. The percentage of similarity is calculated for all the SNP's tested. This percentage is not the overall genetic similarity, but only the similarity in polymorphisms.
The human genome is made of 3,000 million base pairs, there are approximately 15 million SNP's. This leaves plenty of opportunities for genetic variations. At present 23andMe only tests 0.6 million SNP's, so the genetic distance between two persons is yet completely reliable, but it gives a good idea because genes are inherited in blocks (segments).
The first case of two parents being homozygous with the opposite sequence is actually quite rare. It is more likely to happen when the parents come from different ethnic/racial group (e.g. European and Chinese).
The more a population has been mixed over time, the higher its heterozygosity will be. Y-DNA haplogroups can give clues on the level of population admixture. For instance, a population with a lot of haplogroup diversity like Turkey will have more heterozygous people in general than a place like Ireland, where haplogroups are more homogeneous.
China, Korea and Japan are more homozygous than European populations, probably because they expanded from a smaller or more homogeneous Paleolithic population. Consequently, a random Japanese will be genetically closer to a Korean or Chinese than any two random Europeans are from each others. Even two unrelated Irish people will be further apart than a Japanese from a Chinese. This is surely why East Asians look more similar physically and have managed to develop such homogeneous societies (in comparison to Europe's strong individualism).